Contextual language interpretation using large language models

Abstract

One example method includes receiving speech input from a user and obtaining a representation of the speech input. The method further includes adjusting the representation by adding an indication to at least one entity within the representation to determine if disambiguation is needed, and providing the adjusted representation to a language model. The method also includes modifying the adjusted representation by the language model to determine if modification is needed, and providing an output based on the modified and adjusted representations.

Description

[0001] Cross-references to related applications

[0002] This application claims priority to U.S. Patent Application No. 19 / 025,974, filed January 16, 2025, entitled "CONTEXTUAL SPEECH INTERPRETATION USING LARGE LANGUAGE MODELS"; U.S. Provisional Patent Application No. 63 / 665,943, filed June 28, 2024, entitled "CONTEXTUAL SPEECH INTERPRETATION USING LARGE LANGUAGE MODELS"; and U.S. Provisional Patent Application No. 63 / 575,403, filed April 5, 2024, entitled "CONTEXTUAL SPEECH INTERPRETATION USING LARGE LANGUAGE MODELS". The entire contents of each of these patent applications are incorporated herein by reference. Technical Field

[0003] This application relates in general to incorporating large language models into intelligent automated assistant systems, and more specifically to contextual speech interpretation using large language models. Background Technology

[0004] Intelligent automated assistants (or digital assistants) provide a beneficial interface between human users and electronic devices. Such assistants allow users to interact with devices or systems using natural language in spoken and / or textual form. For example, a user can provide verbal input containing their request to a digital assistant running on an electronic device. The digital assistant can interpret the user's intent from this verbal input and act it out as a task. The task can then be performed by executing one or more services of the electronic device, and relevant output in response to the user's request can be returned to the user. Summary of the Invention

[0005] This paper discloses example methods. One example method includes receiving speech input from a user and obtaining a representation of the speech input. The method further includes adjusting the representation by adding an indication to at least one entity within the representation to determine if disambiguation is needed, and providing the adjusted representation to a language model. The method further includes modifying the adjusted representation by the language model to determine if modification is needed, and providing an output based on the modified and adjusted representations.

[0006] This document describes an example non-transitory computer-readable storage medium. An example non-transitory computer-readable storage medium stores one or more programs configured to be executed by one or more processors of a first electronic device, the one or more programs including instructions for receiving speech input from a user and obtaining a representation of the speech input. These instructions also include instructions for determining that at least one entity within the representation needs disambiguation. The representation is adjusted by adding instructions to at least one entity, and the adjusted representation is provided to the language model. These instructions also include modifications based on the determination that the adjusted representation needs to be modified: the adjusted representation is modified by the language model, and output is provided based on the modified and adjusted representation.

[0007] This document describes an example electronic device. An example electronic device includes one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors. The one or more programs include instructions for receiving speech input from a user and obtaining a representation of the speech input. These instructions further include, based on determining that at least one entity within the representation needs disambiguation: adjusting the representation by adding an instruction to at least one entity, and providing the adjusted representation to a language model. These instructions further include, based on determining that the adjusted representation needs modification: modifying the adjusted representation by the language model, and providing output based on the modified and adjusted representation.

[0008] Contextual speech interpretation using large language models (LLMs) provides an enhanced interface for user-device interaction. Specifically, combining LLM techniques with a reference resolution system provides an improved system that leverages the strengths of each component. For example, a reference resolution system can efficiently identify spatial references or otherwise resolve ambiguous terms based on the displayed content or other device context. The use of LLMs provides robust correction and modification capabilities based on training on specific datasets. As an example, LLMs efficiently identify linguistic errors or other features in speech input that might otherwise be difficult for existing digital assistant systems to handle. Therefore, these improvements reduce power consumption and extend device battery life by enabling users to use computer systems faster and more efficiently. Attached Figure Description

[0009] Figure 1 These are block diagrams illustrating various examples of systems and environments used to implement digital assistants.

[0010] Figure 2A This is a block diagram illustrating a portable multi-functional device that implements the client-side portion of a digital assistant according to various examples.

[0011] Figure 2B This is a block diagram illustrating exemplary components for event handling based on various examples.

[0012] Figure 3 Portable multi-functional devices that implement the client-side portion of a digital assistant, based on various examples, are illustrated.

[0013] Figure 4 This is a block diagram of an exemplary multifunctional device with a display and a touch-sensitive surface, based on various examples.

[0014] Figure 5A Examples of user interfaces for menus on portable multi-functional devices, based on various examples, are shown.

[0015] Figure 5B Exemplary user interfaces of multifunctional devices having a touch-sensitive surface separate from the display are illustrated according to various examples.

[0016] Figure 6A Examples of personal electronic devices are shown, based on various examples.

[0017] Figure 6B This is a block diagram illustrating various examples of personal electronic devices.

[0018] Figure 7A It is a block diagram illustrating a digital assistant system or its server portion according to various examples.

[0019] Figure 7B Examples are shown based on various examples. Figure 7A The digital assistant's functions are shown in the image.

[0020] Figure 7C Examples are provided for a portion of the knowledge ontology based on various examples.

[0021] Figure 8 Computer systems based on various examples are illustrated.

[0022] Figure 9 Various components of an exemplary basic system, including a basic model, are illustrated according to various examples.

[0023] Figures 10A to 10B Various systems for contextual speech interpretation using large language models are illustrated, based on a variety of examples.

[0024] Figures 11A to 11B Various processes for contextual speech interpretation using large language models are illustrated based on various examples. Detailed Implementation

[0025] As described in this paper, incorporating large language models (LLMs) into digital assistant systems can enhance the capabilities of digital assistants in various ways, such as by improving how they handle natural language input that includes commands with non-fluent, corrective, or otherwise ambiguous meanings.

[0026] The accompanying drawings are referenced in the following description of the examples, which illustrate specific examples that can be practiced by way of illustration. It should be understood that other examples can be used and structural changes can be made without departing from the scope of the individual examples.

[0027] Furthermore, in a method described herein where one or more steps depend on the satisfaction of one or more conditions, it should be understood that the method may be repeated in multiple repetitions such that, during the repetitions, all conditions determining the steps in the method are satisfied in different repetitions of the method. For example, if the method requires performing a first step (if the conditions are satisfied) and a second step (if the conditions are not satisfied), those skilled in the art will know that the stated steps are repeated until both conditions are satisfied and not satisfied (in no particular order). Thus, a method described as having one or more steps depending on the satisfaction of one or more conditions can be rewritten as a method that repeats until each condition described in the method is satisfied. However, this does not require the system or computer-readable medium to declare that the system or computer-readable medium contains instructions for performing discretionary operations based on the satisfaction of the corresponding one or more conditions, and thus to determine whether possible conditions have been satisfied without explicitly repeating the steps of the method until all conditions determining the steps in the method are satisfied. Those skilled in the art will also understand that, similar to a method having discretionary steps, a system or computer-readable storage medium may repeat the steps of the method multiple times as needed to ensure that all discretionary steps have been performed.

[0028] Although the following description uses the terms "first," "second," etc., to describe various elements, these elements should not be limited by the terms. These terms are merely used to distinguish one element from another. For example, without departing from the various examples described, a first input may be referred to as a second input, and similarly, a second input may be referred to as a first input. Both the first and second inputs are inputs, and in some cases, they are separate and distinct inputs.

[0029] The terminology used in the description of the various examples described herein is for the purpose of describing particular examples only and is not intended to be limiting. As used in the description of the various described examples and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context expressly indicates otherwise. It will also be understood that the term “and / or” as used herein refers to and covers any and all possible combinations of one or more of the associated listed items. It will also be understood that the terms “comprising” and / or “including” as used in this specification specify the presence of the stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0030] Depending on the context, the term "if" can be interpreted as meaning "when," "at," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrases "if it is determined that..." or "if [the stated condition or event] is detected" can be interpreted as meaning "when it is determined that..." or "in response to determination that..." or "when [the stated condition or event] is detected," or "in response to detection of [the stated condition or event]."

[0031] This document describes implementations of electronic devices, user interfaces for such devices, and associated processes for using such devices. In some implementations, the device is a portable communication device that also includes other functions, such as PDA and / or music player functions, such as a mobile phone or a head-mounted device. Exemplary implementations of portable multi-functional devices include, but are not limited to, the Apple Watch from Apple Inc. of Cupertino, California. ® Devices, iPhone ® Devices, iPod Touch ® Devices and iPads ®Device. Alternatively, other portable electronic devices may be used, such as laptops or tablets with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads). It should also be understood that in some embodiments, the device is not a portable communication device, but a desktop computer with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads). In some embodiments, the electronic device is a computer system that communicates with a display generating component (e.g., via wireless or wired communication). The display generating component is configured to provide visual output, such as display via a CRT display, via an LED display, or via image projection. In some embodiments, the display generating component is integrated with the computer system. In some embodiments, the display generating component is separate from the computer system. As used herein, “display” content includes displaying content (e.g., video data rendered or decoded by display controller 256) by sending data (e.g., image data or video data) to an integrated or external display generating component via a wired or wireless connection to visually generate content.

[0032] 1. System and Environment

[0033] Figure 1 A block diagram of system 100 according to various examples is illustrated. In some examples, system 100 implements a digital assistant. The terms "digital assistant," "virtual assistant," "intelligent automated assistant," or "automated digital assistant" refer to any information processing system that interprets natural language input in spoken and / or textual form to infer user intent and performs actions based on the inferred user intent. For example, to act on an inferred user intent, the system performs one or more of the following steps: identifying a task flow with steps and parameters designed to achieve the inferred user intent; inputting a specific request into the task flow based on the inferred user intent; executing the task flow by invoking a program, method, service, API, etc.; and generating an output response to the user in an audible (e.g., verbal) and / or visual form.

[0034] Specifically, a digital assistant can accept user requests that are at least partially in the form of natural language commands, requests, statements, narrations, and / or inquiries. Typically, users request informational responses from the digital assistant or the performance of tasks. A satisfactory response to a user request includes providing the requested informational response, performing the requested task, or a combination of both. For example, a user asks a digital assistant a question such as, “Where am I now?” Based on the user’s current location, the digital assistant answers, “You are near the West Exit of Central Park.” The user also requests to perform a task, such as, “Please invite my friends to my girlfriend’s birthday party next week.” In response, the digital assistant can confirm the request by saying “Okay, right away,” and then send the appropriate calendar invitations to each of the user’s friends listed in the user’s electronic address book. During the performance of a requested task, the digital assistant sometimes interacts with the user in a continuous dialogue involving multiple exchanges of information over extended periods. Many other ways exist to interact with a digital assistant to request information or perform various tasks. In addition to providing verbal responses and taking programmed actions, digital assistants also provide responses in other forms of video or audio, such as text, alerts, music, video, animation, etc.

[0035] like Figure 1 As shown, in some examples, the digital assistant is implemented according to a client-server model. The digital assistant includes a client-side portion 102 (hereinafter referred to as "DA client 102") executing on user device 104 and a server-side portion 106 (hereinafter referred to as "DA server 106") executing on server system 108. DA client 102 communicates with DA server 106 via one or more networks 110. DA client 102 provides client-side functionality, such as user-oriented input and output processing, and communication with DA server 106. DA server 106 provides server-side functionality for any number of DA clients 102 residing on their respective user devices 104.

[0036] In some examples, DA server 106 includes a client-facing I / O interface 112, one or more processing modules 114, data and models 116, and an I / O interface 118 to external services. The client-facing I / O interface 112 facilitates client-facing input and output processing of DA server 106. One or more processing modules 114 utilize data and models 116 to process verbal input and determine user intent based on natural language input. Furthermore, one or more processing modules 114 perform task execution based on the inferred user intent. In some examples, DA server 106 communicates with external services 120 via network 110 to complete tasks or collect information. The I / O interface 118 to external services facilitates such communication.

[0037] User equipment 104 can be any suitable electronic device. In some examples, user equipment 104 is a portable multi-functional device (e.g., see reference below). Figure 2A The described device 200), multi-functional device (e.g., see reference below) Figure 4 The described device 400) or personal electronic device (e.g., referred to below) Figures 6A to 6B The described device (600) is a portable multi-functional device, for example, a mobile phone that also includes other functions such as a PDA and / or music player. Specific examples of portable multi-functional devices include the Apple Watch from Apple Inc. (Cupertino, California). ® iPhone ® iPod Touch ® and iPad ® Devices. Other examples of portable multifunction devices include, but are not limited to, headphones / headsets, speakers, headsets, and laptop or tablet computers. Additionally, in some examples, user device 104 is a non-portable multifunction device. Specifically, user device 104 is a desktop computer, game console, speaker, television, or set-top box. In some examples, user device 104 includes a touch-sensitive surface (e.g., a touchscreen display and / or touchpad). Furthermore, user device 104 may optionally include one or more other physical user interface devices, such as a physical keyboard, mouse, and / or joystick. Various examples of electronic devices, such as multifunction devices, are described in more detail below.

[0038] Examples of communication networks 110 include local area networks (LANs) and wide area networks (WANs), such as the Internet. Communication network 110 is implemented using any known network protocol, including various wired or wireless protocols such as Ethernet, Universal Serial Bus (USB), FireWire, Global System for Mobile Communications (GSM), Enhanced Data GSM Environment (EDGE), Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Bluetooth, Wi-Fi, Voice over Internet Protocol (VoIP), Wi-MAX, or any other suitable communication protocol.

[0039] Server system 108 is implemented on one or more standalone data processing devices or a distributed computer network. In some examples, server system 108 also utilizes various virtual devices and / or services from third-party service providers (e.g., third-party cloud service providers) to provide potential computing and / or infrastructure resources for server system 108.

[0040] In some examples, user equipment 104 communicates with DA server 106 via a second user equipment 122. The second user equipment 122 is similar to or identical to user equipment 104. For example, the second user equipment 122 is similar to the one described below. Figure 2A , Figure 4 and Figures 6A to 6B The described devices are 200, 400, or 600. User equipment 104 is configured to be communicatively coupled to a second user equipment 122 via a direct communication connection (such as Bluetooth, NFC, BTLE, etc.) or via a wired or wireless network (such as a local Wi-Fi network). In some examples, the second user equipment 122 is configured to act as a proxy between user equipment 104 and DA server 106. For example, a DA client 102 of user equipment 104 is configured to send information (e.g., a user request received at user equipment 104) to DA server 106 via the second user equipment 122. DA server 106 processes the information and returns relevant data (e.g., data content in response to the user request) to user equipment 104 via the second user equipment 122.

[0041] In some examples, user equipment 104 is configured to forward a shortened request for data to a second user equipment 122 to reduce the amount of information sent from user equipment 104. The second user equipment 122 is configured to determine supplementary information to add to the shortened request to generate a complete request to be sent to DA server 106. This system architecture can advantageously allow user equipment 104 (e.g., a watch or similar compact electronic device) with limited communication capabilities and / or limited battery power (e.g., a second user equipment 122 with strong communication capabilities and / or battery power, such as a mobile phone, laptop computer, tablet computer, etc.) acting as a proxy to DA server 106 to access the services provided by DA server 106. Although Figure 1 Only two user devices, 104 and 122, are shown, but it should be understood that in some examples, system 100 includes any number and type of user devices configured to communicate with the DA server system 106 in this agent configuration.

[0042] Although Figure 1 The digital assistant illustrated includes both a client-side component (e.g., DA client 102) and a server-side component (e.g., DA server 106), but in some examples, the digital assistant's functionality is implemented as a standalone application installed on the user's device. Furthermore, the functional division between the client and server components of the digital assistant may vary in different implementations. For example, in some examples, the DA client is a thin client that only provides user-facing input and output processing functions, delegating all other functionalities of the digital assistant to the backend server.

[0043] 2. Electronic equipment

[0044] Now let’s turn our attention to the implementation of electronic devices for the client-side portion of a digital assistant. Figure 2A This is a block diagram illustrating a portable multi-functional device 200 with a touch-sensitive display system 212 according to some embodiments. The touch-sensitive display 212 is sometimes referred to as a “touchscreen” for convenience, and is sometimes referred to as or called a “touch-sensitive display system.” Device 200 includes a memory 202 (which optionally includes one or more computer-readable storage media), a memory controller 222, one or more processing units (CPUs) 220, a peripheral interface 218, RF circuitry 208, audio circuitry 210, a speaker 211, a microphone 213, an input / output (I / O) subsystem 206, other input control devices 216, and an external port 224. Device 200 optionally includes one or more optical sensors 264. Device 200 optionally includes one or more contact strength sensors 265 for detecting the intensity of contact on device 200 (e.g., a touch-sensitive surface, such as the touch-sensitive display system 212 of device 200). Device 200 may optionally include one or more haptic output generators 267 for generating haptic outputs on device 200 (e.g., generating haptic outputs on a touch-sensitive surface such as the touch-sensitive display system 212 of device 200 or the touchpad 455 of device 400). These components may optionally communicate via one or more communication buses or signal lines 203.

[0045] As used in this specification and claims, the term "intensity" of contact on a tactile surface refers to the force or pressure (force per unit area) of a contact (e.g., finger contact) on a tactile surface, or to a substitute (alternative) for the force or pressure of a contact on a tactile surface. The intensity of contact has a range of values ​​that includes at least four different values ​​and more typically hundreds of different values ​​(e.g., at least 256). The intensity of contact may optionally be determined (or measured) using various methods and various sensors or combinations of sensors. For example, one or more force sensors below or adjacent to the tactile surface may optionally be used to measure the force at different points on the tactile surface. In some embodiments, force measurements from multiple force sensors are combined (e.g., weighted average) to determine the estimated contact force. Similarly, the pressure-sensitive tip of a stylus may optionally be used to determine the pressure of the stylus on the tactile surface. Alternatively, the size and / or change of the contact area detected on the touch-sensitive surface, the capacitance and / or change of the touch-sensitive surface adjacent to the contact, and / or the resistance and / or change of the touch-sensitive surface adjacent to the contact may be used as substitutes for the force or pressure of the contact on the touch-sensitive surface. In some embodiments, the substitute measurement of the contact force or pressure is used directly to determine whether an intensity threshold has been exceeded (e.g., the intensity threshold is described in units corresponding to the substitute measurement). In some embodiments, the substitute measurement of the contact force or pressure is converted into an estimated force or pressure, and the estimated force or pressure is used to determine whether an intensity threshold has been exceeded (e.g., the intensity threshold is a pressure threshold measured in units of pressure). Using the intensity of the contact as an attribute of user input allows users to access additional device functions that would otherwise be inaccessible to the user on a space-constrained, scaled-down device used (e.g., on a touch-sensitive display) to display an indication and / or receive user input (e.g., via a touch-sensitive display, touch-sensitive surface, or physical / mechanical controls, such as knobs or buttons).

[0046] As used in this specification and claims, the term "tactile output" refers to a physical displacement of the device relative to a previous position of the device, a physical displacement of a component of the device (e.g., a touch-sensitive surface) relative to another component of the device (e.g., the housing), or a displacement of a component relative to the center of mass of the device, detected by the user using the user's tactile sense. For example, when the device or a component of the device comes into contact with a touch-sensitive surface (e.g., a finger, palm, or other part of the user's hand), the tactile output generated by the physical displacement will be interpreted by the user as a tactile sensation corresponding to a perceived change in the physical characteristics of the device or a component of the device. For example, movement of a touch-sensitive surface (e.g., a touch-sensitive display or touchpad) may optionally be interpreted by the user as a "press-click" or "release-click" on a physically actuated button. In some cases, the user will feel a tactile sensation, such as a "press-click" or "release-click," even when a physically actuated button associated with the touch-sensitive surface, which has been physically pressed (e.g., displaced) by the user's movement, does not move. For example, even when the smoothness of the tactile surface remains unchanged, the movement of the tactile surface can optionally be interpreted or sensed by the user as the "roughness" of the tactile surface. While such interpretations of touch by users will be limited by the individualized sensory perceptions of the user, many sensory perceptions of touch are common to most users. Therefore, when a tactile output is described as corresponding to a specific sensory perception of the user (e.g., "release click", "press click", "roughness"), unless otherwise stated, the generated tactile output corresponds to the physical displacement of the device or its components, which will generate the sensory perception described by a typical (or ordinary) user.

[0047] It should be understood that device 200 is merely an example of a portable multifunctional device, and device 200 may optionally have more or fewer components than those shown, may optionally combine two or more components, or may optionally have different configurations or arrangements of these components. Figure 2A The various components shown are implemented in hardware, software, or a combination of both, including one or more signal processing and / or application-specific integrated circuits.

[0048] Memory 202 includes one or more computer-readable storage media. Computer-readable storage media are, for example, tangible and non-transitory. Memory 202 includes high-speed random access memory and also includes non-volatile memory, such as one or more disk storage devices, flash memory devices, or other non-volatile solid-state memory devices. Memory controller 222 controls other components of device 200 to access memory 202.

[0049] In some examples, the non-transitory computer-readable storage medium of memory 202 is used to store instructions (e.g., for performing aspects of the processes described below) for use by or in conjunction with an instruction execution system, apparatus, or device (such as a computer-based system, a processor-containing system, or other system that can fetch and execute instructions from and from an instruction execution system, apparatus, or device). In other examples, the instructions (e.g., for performing aspects of the processes described below) are stored on a non-transitory computer-readable storage medium (not shown) of server system 108, or partitioned between the non-transitory computer-readable storage medium of memory 202 and the non-transitory computer-readable storage medium of server system 108.

[0050] Peripheral interface 218 is used to couple the device's input and output peripherals to CPU 220 and memory 202. One or more processors 220 run or execute various software programs and / or instruction sets stored in memory 202 to perform various functions of device 200 and process data. In some embodiments, peripheral interface 218, CPU 220, and memory controller 222 are implemented on a single chip, such as chip 204. In some other embodiments, they are implemented on separate chips.

[0051] RF (Radio Frequency) circuit 208 receives and transmits RF signals, also known as electromagnetic signals. RF circuit 208 converts electrical signals into electromagnetic signals / converts electromagnetic signals into electrical signals, and communicates with communication networks and other communication devices via electromagnetic signals. RF circuit 208 optionally includes well-known circuitry for performing these functions, including but not limited to antenna systems, RF transceivers, one or more amplifiers, tuners, one or more oscillators, digital signal processors, codec chipsets, subscriber identity module (SIM) cards, memory, etc. RF circuit 208 optionally communicates wirelessly with networks (such as the Internet (also known as the World Wide Web (WWW)), intranets, and / or wireless networks (such as cellular telephone networks, wireless local area networks (LANs), and / or metropolitan area networks (MANs))) and other devices. RF circuit 208 optionally includes well-known circuitry for detecting near-field communication (NFC) fields, such as via short-range communication radio components. Wireless communication may optionally employ any of a variety of communication standards, protocols, and technologies, including but not limited to Global System for Mobile Communications (GSM), Enhanced Data GSM Environment (EDGE), High-Speed ​​Downlink Packet Access (HSDPA), High-Speed ​​Uplink Packet Access (HSUPA), Evolution, Pure Data (EV-DO), HSPA, HSPA+, Dual-Cell HSPA (DC-HSPDA), Long Term Evolution (LTE), Near Field Communication (NFC), Wideband Code Division Multiple Access (W-CDMA), Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Bluetooth, Bluetooth Low Energy (BTLE), and Wi-Fi (e.g., IEEE 802.11a, IEEE 802.11b, IEEE 802.11g, IEEE 802.11n and / or IEEE 802.11ac), Voice over Internet Protocol (VoIP), Wi-MAX, email protocols (e.g., Internet Messaging Access Protocol (IMAP) and / or Post Office Protocol (POP)), instant messaging (e.g., Extensible Messaging Processing and Presence Protocol (XMPP), Session Initiation Protocol for Instant Messaging and Presence Utilizing Extended Protocol (SIMPLE), Instant Messaging and Presence Service (IMPS)) and / or Short Message Service (SMS), or any other suitable communication protocol, including communication protocols that have not yet been developed as of the date of this document submission.

[0052] Audio circuitry 210, speaker 211, and microphone 213 provide an audio interface between the user and device 200. Audio circuitry 210 receives audio data from peripheral interface 218, converts the audio data into electrical signals, and sends the electrical signals to speaker 211. Speaker 211 converts the electrical signals into sound waves that are audible to humans. Audio circuitry 210 also receives electrical signals converted from sound waves by microphone 213. Audio circuitry 210 converts the electrical signals into audio data and sends the audio data to peripheral interface 218 for processing. The audio data is retrieved from and / or sent to memory 202 and / or RF circuitry 208 via peripheral interface 218. In some embodiments, audio circuitry 210 also includes a headset jack (e.g., ...). Figure 3 (312 in the text). The headset jack provides an interface between the audio circuitry 210 and a removable audio input / output peripheral device, such as an output-only headset or a headset with both outputs (e.g., a mono-ear headset or a binaural headset) and inputs (e.g., a microphone).

[0053] I / O subsystem 206 couples input / output peripherals (such as touchscreen 212 and other input control devices 216) on device 200 to peripheral interface 218. I / O subsystem 206 optionally includes display controller 256, optical sensor controller 258, intensity sensor controller 259, haptic feedback controller 261, and one or more input controllers 260 for other input or control devices. The one or more input controllers 260 receive electrical signals from / transmit electrical signals to the other input control device 216. Other input control devices 216 optionally include physical buttons (e.g., push-buttons, rocker buttons, etc.), dial pads, slide switches, joysticks, click dials, etc. In some alternative embodiments, input controller 260 may optionally be coupled to (or not coupled to) any of the following: keyboard, infrared port, USB port, and pointing device such as mouse. One or more buttons (e.g., ... Figure 3 Optionally, 308 may include increase / decrease buttons for volume control of speaker 211 and / or microphone 213. These one or more buttons may optionally include a push-button (e.g., Figure 3(306 in the original text). In some embodiments, the electronic device is a computer system that communicates with one or more input devices (e.g., via wireless communication, via wired communication). In some embodiments, the one or more input devices include a touch-sensitive surface (e.g., a touchpad, as part of a touch-sensitive display). In some embodiments, the one or more input devices include one or more camera sensors (e.g., one or more optical sensors 264, optical sensors included in eye-tracking device 806 and / or optical sensors included in hand-tracking device 810), for example, for tracking the user's posture (e.g., hand posture and / or air posture) as input. In some embodiments, the one or more input devices are integrated with the computer system. In some embodiments, the one or more input devices are separate from the computer system. In some implementations, the air posture is a posture detected without the user touching the input element that is part of the device (or independently of the input element that is part of the device) and based on the detected movement of a part of the user's body through the air (including movement of the user's body relative to an absolute reference (e.g., the angle of the user's arm relative to the ground or the distance of the user's hand relative to the ground), movement relative to another part of the user's body (e.g., movement of the user's hand relative to the user's shoulder, movement of one of the user's hands relative to the user's other hand, and / or movement of the user's fingers relative to another of the user's fingers or a part of the user's hand), and / or absolute movement of a part of the user's body (e.g., a tapping posture that includes the hand moving a predetermined amount and / or speed in a predetermined posture, or a rocking posture that includes a predetermined speed or amount of rotation of a part of the user's body)).

[0054] A rapid press of the push button disengages the touchscreen 212 from its lock or initiates a process of unlocking the device using gestures on the touchscreen, as described in U.S. Patent Application 11 / 322,549 (U.S. Patent No. 7,657,849), filed December 23, 2005, entitled "Unlocking a Device by Performing Gestures on an Unlock Image," the entire contents of which are incorporated herein by reference. A long press of the push button (e.g., 306) powers the device 200 on or off. The user can customize the functionality of one or more buttons. The touchscreen 212 is used to implement virtual buttons or soft buttons and one or more soft keyboards.

[0055] The touch-sensitive display 212 provides input and output interfaces between the device and the user. The display controller 256 receives electrical signals from and / or transmits electrical signals to the touchscreen 212. The touchscreen 212 displays visual output to the user. Visual output includes graphics, text, icons, video, and any combination thereof (collectively, "graphics"). In some embodiments, some or all of the visual output corresponds to user interface objects.

[0056] Touchscreen 212 has a touch-sensitive surface, sensor, or sensor array that accepts input from a user based on tactile and / or haptic contact. Touchscreen 212 and display controller 256 (along with any associated modules and / or instruction set in memory 202) detect contact on touchscreen 212 (and any movement or interruption of that contact) and translate the detected contact into interaction with user interface objects (e.g., one or more soft keys, icons, web pages, or images) displayed on touchscreen 212. In an exemplary embodiment, the contact point between touchscreen 212 and the user corresponds to the user's finger.

[0057] Touchscreen 212 uses LCD (Liquid Crystal Display) technology, LPD (Light Emitting Polymer Display) technology, or LED (Light Emitting Diode) technology, but other display technologies may be used in other embodiments. Touchscreen 212 and display controller 256 use any of a variety of touch sensing technologies currently known or to be developed thereafter, as well as other proximity sensor arrays or other elements for determining one or more points of contact with touchscreen 212 to detect contact and any movement or interruption. These various touch sensing technologies include, but are not limited to, capacitive, resistive, infrared, and surface acoustic wave technologies. In an exemplary embodiment, projected mutual capacitance sensing technology is used, such as in the iPhone from Apple Inc. (Cupertino, California). ® and iPod Touch ® The technology used.

[0058] In some embodiments, the touchscreen 212's touch-sensitive display is similar to the multi-touch panel described in the following U.S. patents: 6,323,846 (Westerman et al.), 6,570,557 (Westerman et al.), and / or 6,677,932 (Westerman) and / or U.S. Patent Publication 2002 / 0015024A1, all of which are incorporated herein by reference in their entirety. However, the touchscreen 212 displays visual output from the device 200, while the touch-sensitive panel does not provide visual output.

[0059] The touch-sensitive display in some embodiments of the touchscreen 212 is described in the following applications: (1) U.S. Patent Application No. 11 / 381,313, filed May 2, 2006, “Multipoint Touch Surface Controller”; (2) U.S. Patent Application No. 10 / 840,862, filed May 6, 2004, “Multipoint Touchscreen”; (3) U.S. Patent Application No. 10 / 903,964, filed July 30, 2004, “Gestures For Touch Sensitive Input Devices”; (4) U.S. Patent Application No. 11 / 048,264, filed January 31, 2005, “Gestures For Touch Sensitive Input Devices”; (5) U.S. Patent Application No. 11 / 038,590, filed January 18, 2005, “Mode-Based Graphical User Interfaces For Touch Sensitive Input Devices”; (6) U.S. Patent Application No. 11 / 228,758, filed September 16, 2005, “Virtual Input Device Placement On A Touch Screen User Interface”; (7) U.S. Patent Application No. 11 / 228,700, filed September 16, 2005, “Operation Of A Computer With A Touch Screen Interface”; (8) U.S. Patent Application No. 11 / 228,737, filed September 16, 2005, “Activating Virtual Keys Of A Touch-Screen Virtual Keyboard”; and (9) U.S. Patent Application No. 11 / 367,749, filed March 3, 2006, “Multi-Functional Hand-Held Device”. The full text of all these applications is incorporated herein by reference.

[0060] Touchscreen 212 has a video resolution of over 100 dpi, for example. In some embodiments, the touchscreen has a video resolution of approximately 160 dpi. The user interacts with touchscreen 212 using any suitable object or accessory such as a stylus, finger, etc. In some embodiments, the user interface is designed to function primarily through finger-based contact and gestures, which may be less precise than stylus-based input due to the larger contact area of ​​a finger on the touchscreen. In some embodiments, the device translates coarse finger-based input into precise pointer / cursor positioning or commands for performing the user-desired actions.

[0061] In some embodiments, in addition to the touchscreen, device 200 also includes a touchpad (not shown) for activating or deactivating specific functions. In some embodiments, the touchpad is a touch-sensitive area of ​​the device that, unlike the touchscreen, does not display visual output. The touchpad is a touch-sensitive surface separate from the touchscreen 212, or an extension of the touch-sensitive surface formed by the touchscreen.

[0062] The device 200 also includes a power system 262 for supplying power to various components. The power system 262 includes a power management system, one or more power sources (e.g., batteries, alternating current (AC)), a recharging system, a power fault detection circuit, a power converter or inverter, a power status indicator (e.g., light-emitting diodes (LEDs)), and any other components associated with the generation, management, and distribution of power in the portable device.

[0063] The device 200 also includes one or more optical sensors 264. Figure 2AAn optical sensor 264 is shown coupled to an optical sensor controller 258 in the I / O subsystem 206. In some embodiments, the optical sensor 264 includes a charge-coupled device (CCD) or a complementary metal-oxide-semiconductor (CMOS) phototransistor. In some embodiments, the optical sensor 264 receives light projected through one or more lenses from the environment and converts the light into data representing an image. In some embodiments, the optical sensor 264 includes a depth camera sensor that receives data from the environment to create a three-dimensional model of objects within a scene from a specific viewpoint. In some embodiments, in conjunction with an imaging module 243 (also referred to as a camera module), the optical sensor 264 captures still images or video. In some embodiments, in conjunction with the imaging module 243, the optical sensor 264 is used to determine depth maps of different portions of the captured image. In some embodiments, the optical sensor is located on the rear of the device 200, facing away from a touchscreen display 212 on the front of the device, such that the touchscreen display is used as a viewfinder for still image and / or video image acquisition. In some implementations, the optical sensor is located on the front of the device, enabling the acquisition of an image of the user for use in the video conference while the user views other video conference participants on a touchscreen display. In some implementations, the positioning of the optical sensor 264 can be changed by the user (e.g., by rotating the lens and sensor within the device housing), allowing a single optical sensor 264 to be used in conjunction with the touchscreen display for both video conferencing and still image and / or video image acquisition.

[0064] The device 200 may optionally also include one or more contact strength sensors 265. Figure 2A A contact strength sensor 265 is shown coupled to a strength sensor controller 259 in I / O subsystem 206. The contact strength sensor 265 optionally includes one or more piezoresistive strain gauges, capacitive force sensors, electro-force sensors, piezoelectric sensors, optical force sensors, capacitive touch-sensitive surfaces, or other strength sensors (e.g., sensors for measuring the force (or pressure) of contact on a touch-sensitive surface). The contact strength sensor 265 receives contact strength information (e.g., pressure information or a substitute for pressure information) from the environment. In some embodiments, at least one contact strength sensor is arranged juxtaposed with or adjacent to a touch-sensitive surface (e.g., touch-sensitive display system 212). In some embodiments, at least one contact strength sensor is located on the rear of device 200, opposite to the touchscreen display 212 located on the front of device 200.

[0065] The device 200 also includes one or more proximity sensors 266. Figure 2AA proximity sensor 266 coupled to a peripheral device interface 218 is shown. Alternatively, the proximity sensor 266 is coupled to an input controller 260 in an I / O subsystem 206. The proximity sensor 266 performs as described in the following U.S. patent applications numbered: 11 / 241,839, entitled "Proximity Detector In Handheld Device"; 11 / 240,788, entitled "Proximity Detector In Handheld Device"; 11 / 620,702, entitled "Using Ambient Light Sensor To Augment Proximity Sensor Output"; 11 / 586,862, entitled "Automated Response To And Sensing Of User Activity In Portable Devices"; and 11 / 638,251, entitled "Methods And Systems For Automatic Configuration Of Peripherals", the entire contents of which are incorporated herein by reference. In some implementations, when the multifunction device is placed near the user's ear (e.g., when the user is making a phone call), the proximity sensor is turned off and the touchscreen 212 is disabled.

[0066] The device 200 may optionally also include one or more haptic output generators 267. Figure 2A A haptic output generator coupled to a haptic feedback controller 261 in I / O subsystem 206 is shown. The haptic output generator 267 optionally includes one or more electroacoustic devices, such as speakers or other audio components, and / or electromechanical devices that convert energy into linear motion, such as motors, solenoids, electroactive polymers, piezoelectric actuators, electrostatic actuators, or other haptic output generating components (e.g., components that convert electrical signals into haptic outputs on the device). A contact intensity sensor 265 receives haptic feedback generation instructions from a haptic feedback module 233 and generates a haptic output on device 200 that can be felt by a user of device 200. In some embodiments, at least one haptic output generator is juxtaposed or adjacent to a haptic surface (e.g., haptic display system 212) and optionally generates the haptic output by moving the haptic surface vertically (e.g., in / outward from the surface of device 200) or laterally (e.g., back and forth in the same plane as the surface of device 200). In some implementations, at least one haptic output generator sensor is located on the rear of the device 200, opposite to the touchscreen display 212 located on the front of the device 200.

[0067] The device 200 also includes one or more accelerometers 268. Figure 2A An accelerometer 268 coupled to a peripheral device interface 218 is shown. Alternatively, the accelerometer 268 is coupled to an input controller 260 in an I / O subsystem 206. The accelerometer 268 performs as described in the following U.S. patent publications: U.S. Patent Publication No. 20050190059, “Acceleration-based Theft Detection System for Portable Electronic Devices” and U.S. Patent Publication No. 20060017692, “Methods and Apparatuses For Operating A Portable Device Based On An Accelerometer,” the entire contents of which are incorporated herein by reference. In some embodiments, information is displayed on a touchscreen display in portrait or landscape view based on analysis of data received from one or more accelerometers. The device 200 may optionally include, in addition to the accelerometer 268, a magnetometer (not shown) and a GPS (or GLONASS or other global navigation system) receiver (not shown) for acquiring information about the location and orientation (e.g., portrait or landscape) of the device 200.

[0068] In some embodiments, software components stored in memory 202 include an operating system 226, a communication module (or instruction set) 228, a contact / motion module (or instruction set) 230, a graphics module (or instruction set) 232, a text input module (or instruction set) 234, a Global Positioning System (GPS) module (or instruction set) 235, a digital assistant client module 229, and an application (or instruction set) 236. Furthermore, memory 202 stores data and models, such as user data and models 231. Additionally, in some embodiments, memory 202 ( Figure 2A ) or 470 ( Figure 4 Storage device / global internal state 257, such as Figure 2A and Figure 4 As shown in the diagram. Device / global internal state 257 includes one or more of the following: active application state, which indicates which applications (if any) are currently active; display state, which indicates what applications, views or other information occupy various areas of the touchscreen display 212; sensor state, which includes information obtained from various sensors and input control devices 216 of the device; and position information relating to the device's position and / or height.

[0069] The operating system 226 (e.g., Darwin, RTXC, LINUX, UNIX, OS X, iOS, WINDOWS, or embedded operating systems such as VxWorks) includes various software components and / or drivers for controlling and managing general system tasks (e.g., memory management, storage device control, power management, etc.) and facilitates communication between various hardware and software components.

[0070] The communication module 228 facilitates communication with other devices via one or more external ports 224 and includes various software components for processing data received by the RF circuitry 208 and / or the external ports 224. The external ports 224 (e.g., Universal Serial Bus (USB), FireWire, etc.) are adapted to be directly coupled to other devices or indirectly coupled via a network (e.g., the Internet, Wireless LAN, etc.). In some embodiments, the external port is for use with an iPod. ® (Trademark of Apple Inc.) The same or similar and / or compatible multi-pin (e.g., 30-pin) connectors used in Apple Inc. devices.

[0071] The contact / motion module 230 optionally detects contact with the touchscreen 212 (in conjunction with the display controller 256) and other touch-sensitive devices (e.g., a touchpad or physical click wheel). The contact / motion module 230 includes various software components for performing various operations related to contact detection, such as determining whether a contact has occurred (e.g., detecting a finger press event), determining the contact intensity (e.g., the force or pressure of the contact, or an alternative to force or pressure), determining whether there is movement of the contact and tracking movement on the touch-sensitive surface (e.g., detecting one or more finger drag events), and determining whether the contact has stopped (e.g., detecting a finger lift event or a contact break). The contact / motion module 230 receives contact data from the touch-sensitive surface. Determining the movement of the contact point optionally includes determining the rate (magnitude), velocity (magnitude and direction), and / or acceleration (change in magnitude and / or direction) of the contact point, the movement of which is represented by a series of contact data. These operations can optionally be applied to single-point contact (e.g., single-finger contact) or simultaneous multi-point contact (e.g., "multi-touch" / multiple-finger contact). In some implementations, the contact / motion module 230 and the display controller 256 detect contact on the touchpad.

[0072] In some implementations, the contact / motion module 230 uses a set of one or more intensity thresholds to determine whether an operation has been performed by the user (e.g., determining whether the user has “clicked” an icon). In some implementations, at least a subset of the intensity thresholds is determined based on software parameters (e.g., the intensity thresholds are not determined by the activation threshold of a specific physical actuator and can be adjusted without changing the physical hardware of the device 200). For example, the mouse “click” threshold of a touchpad or touchscreen can be set to any threshold in a wide range of predefined thresholds without changing the touchpad or touchscreen display hardware. Additionally, in some specific implementations, the user of the device is provided with software settings for adjusting one or more intensity thresholds in a set of intensity thresholds (e.g., by adjusting the individual intensity thresholds and / or by adjusting multiple intensity thresholds at once using system-level clicks on the “intensity” parameter).

[0073] The touch / motion module 230 optionally detects gesture input performed by the user. Different gestures on a touch-sensitive surface have different contact patterns (e.g., different movements, timings, and / or intensities of the detected contact). Therefore, gestures can optionally be detected by detecting specific contact patterns. For example, detecting a finger tap gesture includes: detecting a finger press event, and then detecting a finger lift-off (lift-away) event at the same (or substantially the same) location as the finger press event (e.g., at the location of an icon). As another example, detecting a finger swipe gesture on a touch-sensitive surface includes: detecting a finger press event, then detecting one or more finger drag events, and subsequently detecting a finger lift-off (lift-away) event.

[0074] The graphics module 232 includes various known software components for rendering and displaying graphics on the touchscreen 212 or other displays, including components for altering the visual impact of the displayed graphics (e.g., brightness, transparency, saturation, contrast, or other visual properties). As used herein, the term "graphics" includes any object that can be displayed to a user, including but not limited to text, web pages, icons (such as user interface objects including soft keys), digital images, videos, animations, and so on.

[0075] In some implementations, the graphics module 232 stores data representing the graphics to be used. Each graphic is optionally assigned a corresponding code. The graphics module 232 receives one or more codes from applications, etc., to specify the graphics to be displayed, and also receives coordinate data and other graphic attribute data if necessary, and then generates screen image data to output to the display controller 256.

[0076] The haptic feedback module 233 includes various software components for generating instructions which are used by the haptic output generator 267 to generate haptic output at one or more locations on the device 200 in response to user interaction with the device 200.

[0077] In some examples, the text input module 234, which is a component of the graphics module 232, provides a soft keyboard for entering text in various applications, such as the contact module 237, email client module 240, IM module 241, browser module 247, and any other application that requires text input.

[0078] GPS module 235 determines the location of the device and provides that information for use in various applications (e.g., to telephone module 238 for use in location-based dialing; to camera module 243 as image / video metadata; and to applications that provide location-based services, such as weather widgets, local yellow pages widgets, and map / navigation widgets).

[0079] The digital assistant client module 229 includes various client-side digital assistant commands for providing client-side functionality of the digital assistant. For example, the digital assistant client module 229 can accept voice input (e.g., speech input), text input, touch input, and / or gesture input through various user interfaces of the portable multifunction device 200 (e.g., microphone 213, accelerometer 268, touch-sensitive display system 212, optical sensor 264, other input control devices 216, etc.). The digital assistant client module 229 can also provide audio output (e.g., speech output), visual output, and / or tactile output through various output interfaces of the portable multifunction device 200 (e.g., speaker 211, touch-sensitive display system 212, haptic output generator 267, etc.). For example, the output is provided as voice, sound, alarm, text message, menu, graphics, video, animation, vibration, and / or a combination of two or more of the above. During operation, the digital assistant client module 229 uses RF circuitry 208 to communicate with the DA server 106.

[0080] User data and models 231 include various data associated with the user (e.g., user-specific vocabulary data, user preference data, user-specified name pronunciations, data from the user's electronic address book, to-do lists, shopping lists, etc.) to provide client-side functionality for the digital assistant. Furthermore, user data and models 231 include various models for processing user input and determining user intent (e.g., speech recognition models, statistical language models, natural language processing models, knowledge ontology, task flow models, service models, etc.).

[0081] In some examples, the digital assistant client module 229 utilizes various sensors, subsystems, and peripherals of the portable multifunction device 200 to collect additional information from the surrounding environment of the portable multifunction device 200 to establish a context associated with the user, the current user interaction, and / or the current user input. In some examples, the digital assistant client module 229 provides the contextual information, or a subset thereof, along with the user input to the DA server 106 to help infer the user's intent. In some examples, the digital assistant also uses the contextual information to determine how to prepare output and deliver that output to the user. This contextual information is referred to as contextual data.

[0082] In some examples, the contextual information accompanying user input includes sensor information such as lighting, ambient noise, ambient temperature, and images or videos of the surrounding environment. In some examples, the contextual information may also include the physical state of the device, such as device orientation, device location, device temperature, power level, speed, acceleration, motion pattern, and cellular signal strength. In some examples, information related to the software state of the DA server 106 (e.g., running process, installed programs, past and current network activity, background services, error logs, resource usage, etc.) and the software state of the portable multifunction device 200 is provided to the DA server 106 as contextual information associated with the user input.

[0083] In some examples, the digital assistant client module 229 selectively provides information (e.g., user data 231) stored on the portable multifunction device 200 in response to a request from the DA server 106. In some examples, the digital assistant client module 229 also elicits additional input from the user via natural language dialogue or other user interface, based on a request from the DA server 106. The digital assistant client module 229 passes the additional input to the DA server 106 to assist the DA server 106 in intent inference and / or fulfilling the user intent expressed in the user request.

[0084] The following text is for reference only. Figures 7A to 7C A more detailed description of the digital assistant is described below. It should be understood that the digital assistant client module 229 may include any number of sub-modules of the digital assistant module 726 described below.

[0085] Application 236 includes the following modules (or instruction sets) or subsets or supersets: Contacts module 237 (sometimes called address book or contact list); Telephone module 238; Video conferencing module 239; Email client module 240; Instant Messaging (IM) module 241; Fitness support module 242; Camera module 243 for still images and / or video images; Image management module 244; Video player module; Music player module; Browser module 247; Calendar module 248; Widget module 249, in some examples including one or more of the following: weather widget 249-1, stock market widget 249-2, calculator widget 249-3, alarm clock widget 249-4, dictionary widget 249-5 and other widgets obtained by the user and widgets created by the user 249-6; Module 250 is used to create user-created widgets 249-6; Search module 251; Video and music player module 252, which combines a video player module and a music player module; Memo module 253; Map module 254; and / or Online video module 255.

[0086] Examples of other applications 236 stored in memory 202 include other word processing applications, other image editing applications, drawing applications, rendering applications, Java-enabled applications, encryption, digital access control, speech recognition, and speech duplication.

[0087] In conjunction with the touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, and text input module 234, the contact module 237 manages the address book or contact list (e.g., stored in the application internal state 292 of the contact module 237 in memory 202 or memory 470), including: adding names to the address book; deleting names from the address book; associating phone numbers, email addresses, physical addresses, or other information with names; associating images with names; categorizing and classifying names; providing phone numbers or email addresses to initiate and / or facilitate communication via the telephone module 238, video conferencing module 239, email client module 240, or IM module 241; and so on.

[0088] Combining RF circuitry 208, audio circuitry 210, speaker 211, microphone 213, touchscreen 212, display controller 256, contact / motion module 230, graphics module 232, and text input module 234, telephone module 238 is used to input character sequences corresponding to telephone numbers, access one or more telephone numbers in contact module 237, modify input telephone numbers, dial corresponding telephone numbers, initiate conversations, and disconnect or hang up when a conversation is completed. As noted above, wireless communication uses any of a variety of communication standards, protocols, and technologies.

[0089] Combining RF circuitry 208, audio circuitry 210, speaker 211, microphone 213, touchscreen 212, display controller 256, optical sensor 264, optical sensor controller 258, contact / motion module 230, graphics module 232, text input module 234, contact module 237, and telephone module 238, video conferencing module 239 includes executable instructions to initiate, conduct, and terminate video conferences between the user and one or more other participants based on user instructions.

[0090] Incorporating RF circuitry 208, touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, and text input module 234, email client module 240 includes executable instructions for creating, sending, receiving, and managing emails in response to user commands. Combined with image management module 244, email client module 240 makes it very easy to create and send emails containing still images or video images captured by camera module 243.

[0091] In conjunction with RF circuitry 208, touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, and text input module 234, the instant messaging module 241 includes executable instructions for performing the following operations: entering a character sequence corresponding to an instant message, modifying previously entered characters, sending a corresponding instant message (e.g., using Short Message Service (SMS) or Multimedia Messaging Service (MMS) protocols for telephone-based instant messaging or using XMPP, SIMPLE, or IMPS for internet-based instant messaging), receiving an instant message, and viewing received instant messages. In some embodiments, the instant messages sent and / or received include graphics, photographs, audio files, video files, and / or other attachments supported as in MMS and / or Enhanced Messaging Service (EMS). As used herein, "instant message" refers to both telephone-based messages (e.g., messages delivered using SMS or MMS) and internet-based messages (e.g., messages delivered using XMPP, SIMPLE, or IMPS).

[0092] Incorporating RF circuitry 208, touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, text input module 234, GPS module 235, map module 254, and music player module, fitness support module 242 includes executable instructions for performing the following operations: creating fitness activities (e.g., with time, distance, and / or calorie burning goals); communicating with fitness sensors (exercise equipment); receiving fitness sensor data; calibrating sensors used to monitor fitness; selecting and playing music for fitness activities; and displaying, storing, and transmitting fitness data.

[0093] In conjunction with the touchscreen 212, display controller 256, optical sensor 264, optical sensor controller 258, contact / motion module 230, graphics module 232, and image management module 244, the camera module 243 includes executable instructions for performing the following operations: capturing still images or videos (including video streams) and storing them in memory 202, modifying the characteristics of still images or videos, or deleting still images or videos from memory 202.

[0094] In conjunction with the touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, text input module 234, and camera module 243, the image management module 244 includes executable instructions for performing operations such as arranging, modifying (e.g., editing) or otherwise manipulating, marking, deleting, presenting (e.g., in a digital slideshow or album), and storing still images and / or video images.

[0095] Incorporating RF circuitry 208, touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, and text input module 234, browser module 247 includes executable instructions for performing the following operations: browsing the Internet according to user instructions, including searching, linking to, receiving, and displaying web pages or portions thereof, as well as linking to attachments and other files on web pages.

[0096] Combining RF circuitry 208, touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, text input module 234, email client module 240, and browser module 247, calendar module 248 includes executable instructions to create, display, modify, and store calendars and associated data (e.g., calendar entries, to-dos, etc.) according to user instructions.

[0097] In conjunction with RF circuitry 208, touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, text input module 234, and browser module 247, widget module 249 is a micro-application that can be downloaded and used by a user (e.g., weather widget 249-1, stock market widget 249-2, calculator widget 249-3, alarm clock widget 249-4, and dictionary widget 249-5) or a user-created micro-application (e.g., user-created widget 249-6). In some embodiments, the widget includes HTML (Hypertext Markup Language) files, CSS (Cascading Style Sheets) files, and JavaScript files. In some embodiments, the widget includes XML (Extensible Markup Language) files and JavaScript files (e.g., Yahoo! widgets).

[0098] Combining RF circuit 208, touch screen 212, display controller 256, touch / motion module 230, graphics module 232, text input module 234 and browser module 247, widget creator module 250 is used by the user to create widgets (e.g., to turn user-specified parts of a webpage into widgets).

[0099] In conjunction with the touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, and text input module 234, the search module 251 includes executable instructions for performing the following operations: searching the memory 202 for text, music, sound, images, videos, and / or other files that match one or more search criteria (e.g., one or more user-specified search terms) according to user instructions.

[0100] Incorporating touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, audio circuitry 210, speaker 211, RF circuitry 208, and browser module 247, the video and music player module 252 includes executable instructions allowing users to download and play back recorded music and other sound files stored in one or more file formats such as MP3 or AAC files, as well as executable instructions for displaying, presenting, or otherwise playing back video (e.g., on touchscreen 212 or on an external display connected via external port 224). In some embodiments, device 200 optionally includes the functionality of an MP3 player such as an iPod (a trademark of Apple Inc.).

[0101] Combining the touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, and text input module 234, the memo module 253 includes executable instructions for creating and managing memos, to-do items, etc., according to user instructions.

[0102] Combining RF circuit 208, touch screen 212, display controller 256, contact / motion module 230, graphics module 232, text input module 234, GPS module 235 and browser module 247, map module 254 is used to receive, display, modify and store maps and data associated with the maps (e.g., driving directions, data related to shops and other points of interest at or near a specific location, and other location-based data) according to user instructions.

[0103] Incorporating touchscreen 212, display controller 256, touch / motion module 230, graphics module 232, audio circuitry 210, speaker 211, RF circuitry 208, text input module 234, email client module 240, and browser module 247, the online video module 255 includes instructions to: allow users to access, browse, receive (e.g., via streaming and / or downloading), play back (e.g., on the touchscreen or on an external display connected via external port 224), send emails with links to specific online videos, and otherwise manage online videos in one or more file formats such as H.264. In some embodiments, an instant messaging module 241 is used instead of the email client module 240 to send links to specific online videos. Additional descriptions of the online video application can be found in U.S. Provisional Patent Application No. 60 / 936,562, filed June 20, 2007, entitled “Portable Multifunction Device, Method, and Graphical User Interface for Playing Online Videos,” and U.S. Patent Application No. 11 / 968,067, filed December 31, 2007, entitled “Portable Multifunction Device, Method, and Graphical User Interface for Playing Online Videos,” the contents of which are incorporated herein by reference in their entirety.

[0104] Each of the modules and applications described above corresponds to an executable set of instructions for performing one or more functions described above and the methods described in this patent application (e.g., computer-implemented methods and other information processing methods described herein). These modules (e.g., instruction sets) need not be implemented as separate software programs, processes, or modules; therefore, various subsets of these modules can be combined or otherwise rearranged in various implementations. For example, a video player module can be combined with a music player module into a single module (e.g., Figure 2A (e.g., video and music player module 252). In some embodiments, memory 202 stores a subset of the aforementioned modules and data structures. Additionally, memory 202 stores additional modules and data structures not described above.

[0105] In some implementations, device 200 is a device whose operation of a predefined set of functions is performed solely via a touchscreen and / or touchpad. By using a touchscreen and / or touchpad as the primary input control device for the operation of device 200, the number of physical input control devices (such as push-buttons, dials, etc.) on device 200 is reduced.

[0106] A predefined set of functions unique to touchscreens and / or touchpads may optionally include navigation between user interfaces. In some implementations, the touchpad, when touched by a user, navigates device 200 from any user interface displayed on device 200 to the main menu, main desktop menu, or root menu. In such implementations, a "menu button" is implemented using a touchpad. In some other implementations, the menu button is a physical push-button or other physical input control device, rather than a touchpad.

[0107] Figure 2B This is a block diagram illustrating exemplary components for event handling according to some embodiments. In some embodiments, memory 202 ( Figure 2A ) or memory 470 ( Figure 4 This includes an event classifier 270 (e.g., in operating system 226) and a corresponding application 236-1 (e.g., any of the aforementioned applications 237 to 251, 255, 480 to 490).

[0108] Event classifier 270 receives event information and determines the application 236-1 and application view 291 of application 236-1 to which the event information should be delivered. Event classifier 270 includes event monitor 271 and event dispatcher module 274. In some embodiments, application 236-1 includes application internal state 292, which indicates one or more current application views displayed on touch-sensitive display 212 when the application is active or running. In some embodiments, device / global internal state 257 is used by event classifier 270 to determine which application(s) is currently active, and application internal state 292 is used by event classifier 270 to determine the application view 291 to which the event information should be delivered.

[0109] In some implementations, the application internal state 292 includes additional information such as one or more of the following: recovery information to be used when the application 236-1 resumes execution, user interface state information indicating that information is being displayed or ready to be displayed by the application 236-1, a state queue for enabling the user to return to the previous state or view of the application 236-1, and a repeat / undo queue for the user's previous actions.

[0110] Event monitor 271 receives event information from peripheral device interface 218. The event information includes information about sub-events (e.g., user touch on touch-sensitive display 212 as part of a multi-touch gesture). Peripheral device interface 218 transmits information it receives from I / O subsystem 206 or sensors such as proximity sensor 266, accelerometer 268, and / or microphone 213 (via audio circuitry 210). The information received by peripheral device interface 218 from I / O subsystem 206 includes information from touch-sensitive display 212 or touch-sensitive surfaces.

[0111] In some implementations, the event monitor 271 sends requests to the peripheral device interface 218 at predetermined intervals. In response, the peripheral device interface 218 sends event information. In other implementations, the peripheral device interface 218 sends event information only when a significant event occurs (e.g., receiving input above a predetermined noise threshold and / or receiving input for a predetermined duration).

[0112] In some implementations, the event classifier 270 also includes a hit view determination module 272 and / or an activity event recognizer determination module 273.

[0113] When the touch-sensitive display 212 displays more than one view, the hit view determination module 272 provides a software process for determining where a sub-event has occurred within one or more views. A view consists of controls and other elements that the user can see on the display.

[0114] Another aspect of the user interface associated with an application is the collection of views, sometimes referred to herein as application views or user interface windows, in which information is displayed and touch-based gestures occur. The application view (of the corresponding application) in which a touch is detected corresponds to a procedural level within the application's procedural hierarchy or view hierarchy. For example, the lowest-level view in which a touch is detected is called the hit view, and the set of events identified as correct input is determined at least in part based on the hit view of the initial touch, which begins based on the touch gesture.

[0115] The hit view determination module 272 receives information related to sub-events based on touch gestures. When an application has multiple views organized in a hierarchical structure, the hit view determination module 272 identifies the hit view as the lowest-level view in the hierarchical structure from which the sub-events should be processed. In most cases, the hit view is the lowest-level view in which the initiating sub-event (e.g., the first sub-event in a sequence of sub-events forming an event or potential event) occurs. Once the hit view is identified by the hit view determination module 272, the hit view typically receives all sub-events related to the same touch or input source to which it was identified as the hit view.

[0116] The activity event recognizer determination module 273 determines which views(s) within the view hierarchy should receive a specific sub-event sequence. In some embodiments, the activity event recognizer determination module 273 determines that only the hit view should receive the specific sub-event sequence. In other embodiments, the activity event recognizer determination module 273 determines that all views including the physical location of the sub-event are actively participating views, and therefore determines that all actively participating views should receive the specific sub-event sequence. In other embodiments, even if the touch sub-event is entirely confined to the area associated with a particular view, higher views in the hierarchy will still remain actively participating views.

[0117] Event assigner module 274 assigns event information to event identifiers (e.g., event identifier 280). In embodiments that include active event identifier determination module 273, event assigner module 274 delivers event information to the event identifier determined by active event identifier determination module 273. In some embodiments, event assigner module 274 stores event information in an event queue, which is retrieved by the corresponding event receiver 282.

[0118] In some implementations, operating system 226 includes event classifier 270. Alternatively, application 236-1 includes event classifier 270. In yet another implementation, event classifier 270 is a separate module or part of another module (such as contact / motion module 230) stored in memory 202.

[0119] In some implementations, application 236-1 includes a plurality of event handlers 290 and one or more application views 291, each of which includes instructions for handling touch events occurring within a corresponding view of the application's user interface. Each application view 291 of application 236-1 includes one or more event recognizers 280. Typically, a corresponding application view 291 includes a plurality of event recognizers 280. In other implementations, one or more of the event recognizers 280 are part of a separate module, such as a user interface toolkit (not shown) or a higher-level object from which application 236-1 inherits methods and other properties. In some implementations, the corresponding event handlers 290 include one or more of the following: a data updater 276, an object updater 277, a GUI updater 278, and / or event data 279 received from an event classifier 270. Event handlers 290 utilize or invoke the data updater 276, the object updater 277, or the GUI updater 278 to update the application's internal state 292. Alternatively, one or more application views in application view 291 include one or more corresponding event handlers 290. Additionally, in some embodiments, one or more of data updater 276, object updater 277, and GUI updater 278 are included in the corresponding application view 291.

[0120] The corresponding event identifier 280 receives event information (e.g., event data 279) from the event classifier 270 and identifies the event based on the event information. The event identifier 280 includes an event receiver 282 and an event comparator 284. In some embodiments, the event identifier 280 also includes at least one subset of metadata 283 and event delivery instructions 288 (which includes sub-event delivery instructions).

[0121] Event receiver 282 receives event information from event classifier 270. The event information includes information about sub-events such as touch or touch movement. Depending on the sub-event, the event information also includes additional information, such as the location of the sub-event. When the sub-event involves touch movement, the event information also includes the rate and direction of the sub-event. In some embodiments, the event includes the device rotating from one orientation to another (e.g., from a portrait orientation to a lateral orientation, or vice versa), and the event information includes corresponding information about the device's current orientation (also referred to as device pose).

[0122] Event comparator 284 compares event information with predefined event or sub-event definitions and determines the event or sub-event based on the comparison, or determines or updates the state of the event or sub-event. In some embodiments, event comparator 284 includes event definition 286. Event definition 286 contains definitions of events (e.g., predefined sequences of sub-events), such as event 1 (287-1), event 2 (287-2), and others. In some embodiments, sub-events in event (287) include, for example, touch start, touch end, touch move, touch cancel, and multi-touch. In one example, event 1 (287-1) is defined as a double-click on a displayed object. For example, a double-click includes a first touch (touch start) of a predetermined duration on the displayed object, a first lift-off of a predetermined duration (touch end), a second touch (touch start) of a predetermined duration on the displayed object, and a second lift-off of a predetermined duration (touch end). In another example, event 2 (287-2) is defined as a drag on a displayed object. For example, dragging includes a touch (or contact) of a predetermined duration on the displayed object, movement of the touch on the touch-sensitive display 212, and lifting of the touch (end of touch). In some embodiments, the event also includes information for one or more associated event handlers 290.

[0123] In some implementations, event definition 287 includes definitions of events for corresponding user interface objects. In some implementations, event comparator 284 performs a hit test to determine which user interface object is associated with the sub-event. For example, in an application view displaying three user interface objects on touch-sensitive display 212, when a touch is detected on touch-sensitive display 212, event comparator 284 performs a hit test to determine which of the three user interface objects is associated with the touch (sub-event). If each displayed object is associated with a corresponding event handler 290, the event comparator uses the result of the hit test to determine which event handler 290 should be activated. For example, event comparator 284 selects the event handler associated with the sub-event and the object that triggered the hit test.

[0124] In some implementations, the definition of the corresponding event (287) also includes a delay action that delays the delivery of event information until it has been determined whether the sub-event sequence actually corresponds to or does not correspond to the event type of the event recognizer.

[0125] When the corresponding event recognizer 280 determines that the sub-event sequence does not match any event in event definition 286, the corresponding event recognizer 280 enters an event impossible, event failed, or event ended state, after which subsequent touch-based gesture-based sub-events are ignored. In this case, other event recognizers (if any) that remain active for the hit view continue to track and process ongoing touch-based gesture-based sub-events.

[0126] In some embodiments, the corresponding event recognizer 280 includes metadata 283 having configurable attributes, flags, and / or lists instructing how the event delivery system should perform sub-event delivery to actively participating event recognizers. In some embodiments, the metadata 283 includes configurable attributes, flags, and / or lists instructing how or how event recognizers can interact with each other. In some embodiments, the metadata 283 includes configurable attributes, flags, and / or lists instructing whether sub-events are delivered to different levels in a view or programmatic hierarchy.

[0127] In some implementations, when one or more specific sub-events of an event are identified, the corresponding event recognizer 280 activates the event handler 290 associated with the event. In some implementations, the corresponding event recognizer 280 delivers event information associated with the event to the event handler 290. Activating the event handler 290 is different from delivering (and deferred delivering) the sub-events to the corresponding hit view. In some implementations, the event recognizer 280 throws a flag associated with the identified event, and the event handler 290 associated with the flag retrieves the flag and performs a predefined process.

[0128] In some implementations, event delivery instruction 288 includes a sub-event delivery instruction that delivers event information about a sub-event without activating an event handler. Instead, the sub-event delivery instruction delivers the event information to an event handler associated with the sub-event sequence or to an actively participating view. The event handler associated with the sub-event sequence or the actively participating view receives the event information and performs a predetermined process.

[0129] In some implementations, data updater 276 creates and updates data used in application 236-1. For example, data updater 276 updates phone numbers used in contact module 237 or stores video files used in video player module. In some implementations, object updater 277 creates and updates objects used in application 236-1. For example, object updater 277 creates new user interface objects or updates the positioning of user interface objects. GUI updater 278 updates the GUI. For example, GUI updater 278 prepares display information and transmits that display information to graphics module 232 for display on a touch-sensitive display.

[0130] In some implementations, event handler 290 includes, or has access to, a data updater 276, an object updater 277, and a GUI updater 278. In some implementations, data updater 276, object updater 277, and GUI updater 278 are included in a single module of the corresponding application 236-1 or application view 291. In other implementations, they are included in two or more software modules.

[0131] It should be understood that the above discussion regarding event handling for user touch on a touch-sensitive display also applies to other forms of user input that utilize input devices to operate the multifunction device 200, and not all user input is initiated on a touchscreen. For example, mouse movement and mouse button presses optionally in conjunction with single or multiple keyboard presses or holds; touch movements on the touchpad, such as taps, drags, scrolls, etc.; stylus input; device movement; verbal commands; detected eye movements; biometric input; and / or any combination thereof may optionally be used as input corresponding to sub-events that define the event to be identified.

[0132] Figure 3A portable multi-functional device 200 with a touchscreen 212 is illustrated according to some embodiments. The touchscreen optionally displays one or more graphics within a user interface (UI) 300. In this embodiment and other embodiments described below, a user can select one or more graphics by gesturing over the graphics, for example, using one or more fingers 302 (not drawn to scale in the illustration) or one or more styluses 303 (not drawn to scale in the illustration). In some embodiments, selection of one or more graphics occurs when the user breaks contact with one or more graphics. In some embodiments, gestures optionally include one or more taps, one or more swipes (from left to right, from right to left, up and / or down), and / or scrolling (from right to left, from left to right, up and / or down) of a finger already in contact with the device 200. In some specific embodiments or in some cases, unintentional contact with a graphic does not select the graphic. For example, a swipe gesture over an application icon may optionally not select the corresponding application when the gesture corresponding to selection is a tap.

[0133] Device 200 also includes one or more physical buttons, such as a "main desktop" or menu button 304. As previously described, menu button 304 is used to navigate to any application 236 of a set of applications running on device 200. Alternatively, in some embodiments, the menu button is implemented as a soft key in a GUI displayed on touchscreen 212.

[0134] In some embodiments, device 200 includes a touchscreen 212, a menu button 304, a push-button 306 for powering on / off and locking the device, a volume control button 308, a SIM card slot 310, a headset jack 312, and a docking / charging external port 224. The push-button 306 may optionally be used to: power on / off the device by pressing the button and holding it in the pressed state for a predefined time interval; lock the device by pressing the button and releasing it before the predefined time interval has elapsed; and / or unlock the device or initiate an unlocking process. In another embodiment, device 200 also accepts voice input via microphone 213 for activating or deactivating certain functions. Device 200 may also optionally include one or more contact strength sensors 265 for detecting the intensity of contact on the touchscreen 212, and / or one or more haptic output generators 267 for generating haptic output for the user of device 200.

[0135] Figure 4This is a block diagram of an exemplary multi-functional device with a display and a touch-sensitive surface according to some embodiments. Device 400 need not be portable. In some embodiments, device 400 is a laptop computer, desktop computer, tablet computer, multimedia player device, navigation device, educational device (such as a children's learning toy), gaming system, or control device (e.g., a home controller or industrial controller). Device 400 typically includes one or more processing units (CPUs) 410, one or more network or other communication interfaces 460, memory 470, and one or more communication buses 420 for interconnecting these components. Communication bus 420 optionally includes circuitry (sometimes referred to as a chipset) that interconnects system components and controls communication between system components. Device 400 includes an input / output (I / O) interface 430 with a display 440, which is typically a touchscreen display. I / O interface 430 may also optionally include a keyboard and / or mouse (or other pointing device) 450 and a touchpad 455, and a haptic output generator 457 for generating haptic output on device 400 (e.g., similar to the reference above). Figure 2A The described tactile output generator 267), sensor 459 (e.g., optical sensor, accelerometer, proximity sensor, touch sensor and / or contact intensity sensor (similar to the one described above)) Figure 2A The described contact strength sensor 265). Memory 470 includes high-speed random access memory, such as DRAM, SRAM, DDR RAM, or other random access solid-state memory devices; and optionally includes non-volatile memory, such as one or more disk storage devices, optical disk storage devices, flash memory devices, or other non-volatile solid-state storage devices. Memory 470 optionally includes one or more storage devices located remotely from CPU 410. In some embodiments, memory 470 stores data with portable multifunction device 200 (…). Figure 2A The memory 470 stores programs, modules, and data structures similar to those in the memory 202 of the portable multifunction device 200, or subsets thereof. Additionally, the memory 470 may optionally store additional programs, modules, and data structures not present in the memory 202 of the portable multifunction device 200. For example, the memory 470 of the device 400 may optionally store a drawing module 480, a rendering module 482, a word processing module 484, a website creation module 486, a disk editing module 488, and / or a spreadsheet module 490, while the portable multifunction device 200 ( Figure 2A The memory 202 may optionally not store these modules.

[0136] Figure 4Each of the aforementioned elements is stored in one or more of the previously mentioned memory devices in some examples. Each of the aforementioned modules corresponds to a set of instructions for performing the functions described above. The aforementioned modules or programs (e.g., instruction sets) need not be implemented as separate software programs, processes, or modules, and therefore various subsets of these modules are combined or otherwise rearranged in various embodiments. In some embodiments, memory 470 stores a subset of the aforementioned modules and data structures. In addition, memory 470 stores additional modules and data structures not described above.

[0137] Now let’s turn our attention to implementations of user interfaces that can be implemented, for example, on a portable multi-functional device 200.

[0138] Figure 5A An exemplary user interface for an application menu on a portable multifunction device 200 according to some embodiments is illustrated. A similar user interface is implemented on device 400. In some embodiments, user interface 500 includes the following elements or a subset or superset thereof: Signal strength indicator 502 for wireless communications (such as cellular signals and Wi-Fi signals); Time 504; Bluetooth indicator 505; Battery status indicator 506; Tray 508 features icons for frequently used applications, such as: o The telephone module 238 has an icon 516 labeled "telephone", which optionally includes an indicator 514 for the number of missed calls or voicemails; o An icon 518 labeled "Mail" in the email client module 240, which optionally includes an indicator 510 for the number of unread emails; o Browser module 247's icon 520 labeled "Browser"; and o The video and music player module 252 (also known as the iPod (Apple Inc. trademark) module 252) is marked with an icon 522 labeled "iPod"; and Icons of other applications, such as: o IM module 241's icon 524 labeled "Message"; o Calendar module 248's icon 526 labeled "Calendar"; o Image management module 244's icon 528 labeled "Photo"; o The icon 530 of the camera module 243 is labeled "camera"; o Icon 532 of the online video module 255, labeled "Online Video"; o Stock Market widget 249-2 with icon 534 labeled "Stock Market"; o Map module 254's icon 536 labeled "map"; o Weather widget 249-1 with icon 538 labeled "Weather"; o Alarm clock widget 249-4 with icon 540 labeled "clock"; o The icon 542 of the fitness support module 242 is labeled "fitness support"; o The icon 544 labeled "Memo" in the Memo module 253; and o A settings icon 546 labeled "Settings" is used to set up an application or module, which provides access to the settings of the device 200 and its various applications 236.

[0139] It should be pointed out that, Figure 5A The illustrated icon labels are merely exemplary. For example, the icon 522 of the video and music player module 252 may optionally be labeled "Music" or "Music Player". Other labels may optionally be used for various application icons. In some embodiments, the label of a particular application icon includes the name of the application corresponding to that particular application icon. In some embodiments, the label of a particular application icon is different from the name of the application corresponding to that particular application icon.

[0140] Figure SB illustrates a touch-sensitive surface 551 (e.g., separate from the display 550 (e.g., touchscreen display 212)). Figure 4 Devices such as tablets or touchpads (455) Figure 4 An exemplary user interface on the device 400. The device 400 may also optionally include one or more contact intensity sensors (e.g., one or more of the sensors 459) for detecting the intensity of contact on the tactile surface 551 and / or one or more tactile output generators 457 for generating tactile output for the user of the device 400.

[0141] While some examples of input on a touchscreen display 212 (which combines a touch-sensitive surface and a display) are given below, in some embodiments, the device detects input on a touch-sensitive surface separate from the display, as shown in FIG. SB. In some embodiments, the touch-sensitive surface (e.g., 551 in FIG. SB) has a main axis (e.g., 552 in FIG. SB) corresponding to a main axis (e.g., 553 in FIG. SB) on the display (e.g., 550). According to these embodiments, the device detects contact with the touch-sensitive surface 551 at locations corresponding to corresponding positions on the display (e.g., in FIG. SB, contact 560 corresponds to 568 and contact 562 corresponds to 570). Thus, when the touch-sensitive surface (e.g., Figure 5B 551 in the middle) and the display of a multi-functional device (e.g., Figure 5B When 550 is separated from the touch-sensitive surface, user input detected by the device (e.g., touches 560 and 562 and their movement) is used by the device to manipulate the user interface on the display. It should be understood that similar methods may be used for other user interfaces described herein.

[0142] Additionally, while the examples below are given primarily with reference to finger input (e.g., finger touch, finger tap, finger swipe), it should be understood that in some implementations, one or more of these finger inputs may be replaced by input from another input device (e.g., mouse-based input or stylus input). For example, a swipe may optionally be replaced by a mouse click (e.g., instead of a touch), followed by movement of the cursor along the path of the swipe (e.g., instead of movement of the touch). Similarly, a tap may optionally be replaced by a mouse click while the cursor is above the position of the tap (e.g., instead of detection of the touch, followed by cessation of touch detection). Likewise, when multiple user inputs are detected simultaneously, it should be understood that multiple computer mice may optionally be used simultaneously, or mouse and finger touch may optionally be used simultaneously.

[0143] Figure 6A An exemplary personal electronic device 600 is illustrated. Device 600 includes a body 602. In some embodiments, device 600 includes components relative to devices 200 and 400 (e.g., Figures 2A to 4The device 600 may contain some or all of the features described herein. In some embodiments, the device 600 has a touch-sensitive display 604, referred to below as a touchscreen 604. Alternatively, or in addition to the touchscreen 604, the device 600 may also have a display and a touch-sensitive surface. Similar to the cases of devices 200 and 400, in some embodiments, the touchscreen 604 (or touch-sensitive surface) has one or more intensity sensors for detecting the intensity of an applied contact (e.g., a touch). The one or more intensity sensors of the touchscreen 604 (or touch-sensitive surface) provide output data representing the intensity of the touch. The user interface of the device 600 responds to touches based on the intensity of the touch, meaning that touches of different intensities may invoke different user interface operations on the device 600.

[0144] Techniques for detecting and processing touch intensity may exist, for example, in the following related applications: International Patent Application Serial No. PCT / US2013 / 040061, filed May 8, 2013, entitled “Device, Method, and Graphical User Interface for Displaying User Interface Objects Corresponding to an Application”, and International Patent Application Serial No. PCT / US2013 / 069483, filed November 11, 2013, entitled “Device, Method, and Graphical User Interface for Transitioning Between Touch Input to Display Output Relationships”, each of which is incorporated herein by reference in its entirety.

[0145] In some embodiments, device 600 has one or more input mechanisms 606 and 608. Input mechanisms 606 and 608, if included, are physical. Examples of physical input mechanisms include push-buttons and rotatable mechanisms. In some embodiments, device 600 has one or more attachment mechanisms. Such attachment mechanisms, if included, allow device 600 to be attached to, for example, hats, glasses, earrings, necklaces, shirts, jackets, bracelets, watch straps, bangles, trousers, belts, shoes, wallets, backpacks, etc. These attachment mechanisms allow a user to wear device 600.

[0146] Figure 6B An exemplary personal electronic device 600 is depicted. In some embodiments, device 600 includes... Figure 2A , Figure 2B and Figure 4Some or all of the components described. Device 600 has a bus 612 that operatively couples I / O portion 614 to one or more computer processors 616 and memory 618. I / O portion 614 is connected to display 604, which may have touch-sensitive component 622 and optionally also has touch intensity-sensitive component 624. Furthermore, I / O portion 614 is connected to communication unit 630 for receiving application and operating system data using Wi-Fi, Bluetooth, near field communication (NFC), cellular and / or other wireless communication technologies. Device 600 includes input mechanisms 606 and / or 608. For example, input mechanism 606 is a rotatable input device or a pressable and rotatable input device. In some examples, input mechanism 608 is a button.

[0147] In some examples, the input mechanism 608 is a microphone. The personal electronic device 600 includes, for example, various sensors such as a GPS sensor 632, an accelerometer 634, an orientation sensor 640 (e.g., a compass), a gyroscope 636, a motion sensor 638, and / or combinations thereof, all of which are operatively connected to the I / O section 614.

[0148] The memory 618 of the personal electronic device 600 is a non-transitory computer-readable storage medium for storing computer-executable instructions that, when executed by one or more computer processors 616, cause the computer processors to perform, for example, the techniques and processes described below. These computer-executable instructions are also stored and / or transported, for example, within any non-transitory computer-readable storage medium, for use by or in conjunction with an instruction execution system, apparatus, or device, such as a computer-based system, a processor-integrated system, or other system capable of retrieving and executing instructions from and from an instruction execution system, apparatus, or device. The personal electronic device 600 is not limited to... Figure 6B It can be the components and configurations, or it can include other components or additional components in a variety of configurations.

[0149] As used herein, the term "power indication" refers, for example, in devices 200, 400, 600, 800 and / or 1000 ( Figure 2A , Figure 4 , Figures 6A to 6B , Figure 8 and Figures 10A to 10B A graphical user interface object displayed on a screen. For example, images (e.g., icons), buttons, and text (e.g., hyperlinks) each constitute a representation.

[0150] As used herein, the term "focus selector" refers to an input element used to indicate the current portion of a user interface with which a user is interacting. In some specific implementations that include a cursor or other positional marker, the cursor acts as a "focus selector," such that when the cursor is over a particular user interface element (e.g., a button, window, slider, or other user interface element), the cursor is positioned on a touch-sensitive surface (e.g., a...). Figure 4 The touchpad 455 or Figure 5B When an input (e.g., a press input) is detected on the touch-sensitive surface 551 of the display, the specific user interface element is adjusted according to the detected input. This applies to touchscreen displays (e.g., those capable of direct interaction with user interface elements on a touchscreen display) that enable direct interaction with user interface elements on the touchscreen display. Figure 2A The touch-sensitive display system 212 or Figure 5A In some embodiments of the touchscreen 212, a touch detected on the touchscreen acts as a "focus selector," such that when input (e.g., a press input by touch) is detected at the location of a particular user interface element (e.g., a button, window, slider, or other user interface element) on the touchscreen display, that particular user interface element is adjusted according to the detected input. In some embodiments, focus moves from one area of ​​the user interface to another without corresponding movement of the cursor or movement of a touch on the touchscreen display (e.g., moving focus from one button to another using tab keys or arrow keys); in these embodiments, the focus selector moves according to the movement of focus between different areas of the user interface. Regardless of the specific form the focus selector takes, the focus selector is typically a user-controlled user interface element (or a touch on the touchscreen display) that conveys the user's expected interaction with the user interface (e.g., by indicating to the device the elements of the user interface that the user expects to interact with). For example, when a press input is detected on a touch-sensitive surface (e.g., a touchpad or touchscreen), the position of the focus selector (e.g., a cursor, touch, or selection box) above the corresponding button will indicate to the user that they expect to activate the corresponding button (rather than other user interface elements shown on the device's display).

[0151] As used in the specification and claims, the term "characteristic strength" of a contact refers to a characteristic of the contact based on one or more strengths of the contact. In some embodiments, the characteristic strength is based on multiple strength samples. The characteristic strength may optionally be based on a predefined number of strength samples or a set of strength samples collected over a predetermined time period (e.g., 0.05 seconds, 0.1 seconds, 0.2 seconds, 0.5 seconds, 1 second, 2 seconds, 5 seconds, 10 seconds) relative to a predefined event (e.g., after contact is detected, before contact is detected to be lifted off, before or after contact begins to move, before contact ends, before or after contact strength is detected to increase, and / or before or after contact strength is detected to decrease). The characteristic strength of the contact may optionally be based on one or more of the following: the maximum value of the contact strength, the mean value of the contact strength, the average value of the contact strength, the value at the top 10% of the contact strength, the half maximum value of the contact strength, the 90% maximum value of the contact strength, etc. In some embodiments, the duration of the contact is used when determining the characteristic strength (e.g., when the characteristic strength is the average value of the contact strength over time). In some implementations, the characteristic intensity is compared to a set of one or more intensity thresholds to determine whether a user has performed an action. For example, the set of one or more intensity thresholds may include a first intensity threshold and a second intensity threshold. In this example, contact with a characteristic intensity not exceeding the first threshold results in a first action, contact with a characteristic intensity exceeding the first intensity threshold but not exceeding the second intensity threshold results in a second action, and contact with a characteristic intensity exceeding the second threshold results in a third action. In some implementations, the comparison between the characteristic intensity and one or more thresholds is used to determine whether to perform one or more actions (e.g., whether to perform the corresponding action or abort performing the corresponding action), rather than to determine whether to perform the first or second action.

[0152] In some implementations, a portion of the gesture is identified for determining the characteristic strength. For example, a touch-sensitive surface receives a series of swipe contacts that transition from a starting position to an ending position, where the contact strength increases. In this example, the characteristic strength of the contact at the ending position is based only on a portion of the series of swipe contacts, rather than the entire swipe contact (e.g., the swipe contact only covers the portion at the ending position). In some implementations, a smoothing algorithm is applied to the strength of the swipe contact before determining its characteristic strength. For example, the smoothing algorithm may optionally include one or more of the following: unweighted moving average smoothing algorithm, triangular smoothing algorithm, median filter smoothing algorithm, and / or exponential smoothing algorithm. In some cases, these smoothing algorithms eliminate narrow spikes or dips in the strength of the swipe contact to achieve the purpose of determining the characteristic strength.

[0153] The intensity of contact on a touch-sensitive surface is characterized relative to one or more intensity thresholds, such as a contact detection intensity threshold, a light press intensity threshold, a deep press intensity threshold, and / or one or more other intensity thresholds. In some embodiments, the light press intensity threshold corresponds to an intensity at which the device performs an operation typically associated with clicking a button on a physical mouse or touchpad. In some embodiments, the deep press intensity threshold corresponds to an intensity at which the device performs an operation different from the operation typically associated with clicking a button on a physical mouse or touchpad. In some embodiments, when a contact with an intensity lower than the light press intensity threshold (e.g., and higher than the nominal contact detection intensity threshold, where contacts lower than the nominal contact detection intensity threshold are no longer detected) is detected, the device will move the focus selector based on the movement of the contact on the touch-sensitive surface without performing the operation associated with the light press intensity threshold or the deep press intensity threshold. Generally, unless otherwise stated, these intensity thresholds are consistent across different groups of user interface figures.

[0154] An increase in contact intensity from below a light press intensity threshold to an intensity between the light press intensity threshold and the deep press intensity threshold is sometimes referred to as a "light press" input. An increase in contact intensity from below a deep press intensity threshold to an intensity above the deep press intensity threshold is sometimes referred to as a "deep press" input. An increase in contact intensity from below a contact detection intensity threshold to an intensity between the contact detection intensity threshold and the light press intensity threshold is sometimes referred to as detecting a contact on the touch surface. A decrease in contact intensity from above a contact detection intensity threshold to an intensity below the contact detection intensity threshold is sometimes referred to as detecting a contact being lifted off the touch surface. In some embodiments, the contact detection intensity threshold is zero. In some embodiments, the contact detection intensity threshold is greater than zero.

[0155] In some embodiments described herein, one or more operations are performed in response to detecting a gesture including a corresponding press input or in response to detecting a corresponding press input performed using a corresponding contact (or multiple contacts), wherein the corresponding press input is detected at least in part based on detecting that the intensity of the contact (or multiple contacts) increases to above a press input intensity threshold. In some embodiments, the corresponding operation is performed in response to detecting that the intensity of the corresponding contact increases to above a press input intensity threshold (e.g., a "downward stroke" of the corresponding press input). In some embodiments, the press input includes the intensity of the corresponding contact increasing to above a press input intensity threshold and the intensity of the contact subsequently decreasing to below the press input intensity threshold, and the corresponding operation is performed in response to detecting that the intensity of the corresponding contact subsequently decreases to below the press input threshold (e.g., an "upward stroke" of the corresponding press input).

[0156] In some implementations, the device employs intensity hysteresis to avoid unintended inputs sometimes referred to as "jitter," wherein the device defines or selects a hysteresis intensity threshold that has a predefined relationship with a press input intensity threshold (e.g., the hysteresis intensity threshold is X intensity units lower than the press input intensity threshold, or the hysteresis intensity threshold is 75%, 90%, or some reasonable percentage of the press input intensity threshold). Therefore, in some implementations, a press input includes an increase in the intensity of the corresponding contact above the press input intensity threshold and a subsequent decrease in the intensity of that contact below the hysteresis intensity threshold corresponding to the press input intensity threshold, and an operation is performed in response to detecting that the intensity of the corresponding contact subsequently decreases below the hysteresis intensity threshold (e.g., the "upstroke" of the corresponding press input). Similarly, in some implementations, a press input is detected only when the device detects that the intensity of the contact increases from an intensity equal to or below a hysteresis intensity threshold to an intensity equal to or above a press input intensity threshold and optionally the intensity of the contact subsequently decreases to an intensity equal to or below the hysteresis intensity, and corresponding operations are performed in response to the detection of a press input (e.g., depending on the environment, the intensity of the contact increases or decreases).

[0157] For ease of explanation, optionally, the description of an operation triggered in response to a press input associated with a press input strength threshold or in response to a gesture including a press input may be provided in response to detecting any of the following conditions: the contact strength increases to above the press input strength threshold, the contact strength increases from below a hysteresis strength threshold to above the press input strength threshold, the contact strength decreases to below the press input strength threshold, and / or the contact strength decreases to below a hysteresis strength threshold corresponding to the press input strength threshold. Additionally, in the example where the operation is described as being performed in response to detecting a decrease in contact strength below the press input strength threshold, the operation may optionally be performed in response to detecting a decrease in contact strength below a hysteresis strength threshold corresponding to and less than the press input strength threshold.

[0158] As used herein, "installed application" refers to a software application that has been downloaded to an electronic device (e.g., device 100, 300, and / or 500) and is ready to be launched on the device (e.g., become open). In some implementations, the downloaded application becomes an installed application using an installer that extracts program portions from the downloaded software package and integrates the extracted portions with the computer system's operating system.

[0159] As used herein, the terms "open application" or "running application" refer to a software application that maintains state information (e.g., as part of device / global internal state 157 and / or application internal state 192). An open or running application can optionally be any of the following types of applications: The active application, which is currently displayed on the screen of the device that is using the application; Background applications (or background processes) that are not currently displayed but whose one or more processes are being handled by one or more processors; and A paused or hibernating application that is not currently running but has state information stored in memory (either volatile or non-volatile) that can be used to resume the application's execution.

[0160] As used herein, the term "closed application" refers to a software application that does not retain state information (e.g., the state information of a closed application is not stored in the device's memory). Therefore, closing an application includes: stopping and / or removing the application's process and removing the application's state information from the device's memory. Generally, opening a second application while the first application is running does not close the first application. When the second application is displayed and the first application stops displaying, the first application becomes a background application.

[0161] As described herein, content is automatically generated by one or more computers in response to a request to generate content. The automatically generated content may optionally be generated on a device (e.g., at least in part by the computer system that received the request to generate content) and / or generated outside the device (e.g., at least in part by one or more nearby computers available via a local network or one or more computers available via the Internet). The automatically generated content may optionally include visual content (e.g., images, graphics, and / or video), audio content, and / or text content.

[0162] In some implementations, novel, automatically generated content produced via one or more artificial intelligence (AI) processes is referred to as generative content (e.g., generative images, generative graphics, generative videos, generative audio, and / or generative text). Generative content is typically generated by an AI process based on prompts provided to that AI process. The AI ​​process typically uses one or more AI models to generate output based on input. The AI ​​process may optionally include one or more preprocessing steps (e.g., adjusting user-provided prompts, creating system-generated prompts, and / or selecting AI models) to adjust the input before it is used by the AI ​​model to generate output. The AI ​​process may optionally include one or more postprocessing steps (e.g., passing the AI ​​model output to different AI models, scaling up, down, cropping, formatting, and / or adding or removing metadata) to adjust the AI ​​model output before it is used for other purposes (e.g., providing it to different software processes for further processing or presenting it to a user, for example, visually or aurally). The AI ​​process that generates generative content is sometimes referred to as a generative AI process.

[0163] Hints used to generate generative content may include one or more of the following: one or more words (e.g., natural language hindrances in written or spoken language), one or more images, one or more pictures, and / or one or more videos. AI processes may include machine learning models, which may include neural networks. Neural networks may include transformer-based deep neural networks, such as Large Language Models (LLMs). A generative pre-trained transformer model is an LLM that can efficiently generate novel generative content based on hindrances. Some AI processes use hindrances that include text to generate diverse generative text, generative audio content, and / or generative visual content. Some AI processes use hindrances that include visual content and / or audio content to generate generative text (e.g., transcriptions of audio and / or descriptions of visual content). Some multimodal AI processes use hindrances that include multiple types of content (e.g., text, images, audio, video, and / or other sensor data) to generate generative content. Hints may sometimes also include values ​​for one or more parameters indicating the importance of the various parts of the hindrance. Some tips include a structured set of instructions that the AI ​​process can understand, including wording, specified style, relevant context (e.g., starting content and / or one or more examples) and / or the role of the AI ​​process.

[0164] Generative content is typically based on a cue, but it is not deterministically selected from pre-generated content; rather, it is generated using the cue as a starting point. In some implementations, pre-existing content (e.g., audio, text, and / or visual content) is used as part of the cue for creating the generative content (e.g., pre-existing content is used as a starting point for creating the generative content). For example, the cue may request that a piece of text be summarized or rewritten in a different tone, and the output will be the generative text summarized or written in a different tone. Similarly, the cue may request that visual content be modified to include or exclude content specified by the cue (e.g., removing identified features from the visual content, adding features described in the cue to the visual content, changing the visual style of the visual content, and / or creating additional visual elements based on the visual content beyond its spatial or temporal boundaries). In some implementations, a random or pseudo-random seed is used as part of the cue for creating the generative content (e.g., random or pseudo-random seed content is used as a starting point for creating the generative content). For example, when generating images from a diffusion model, random noise patterns are iteratively denoised based on the cue to generate images based on the cue. While this article has described specific types of AI processes, it should be understood that a variety of different AI processes can be used to generate generative content based on prompts.

[0165] 3. Digital Assistant System

[0166] Figure 7A Block diagrams of digital assistant systems 700 according to various examples are illustrated. In some examples, the digital assistant system 700 is implemented on a standalone computer system. In some examples, the digital assistant system 700 is distributed across multiple computers. In some examples, some modules and functions of the digital assistant are divided into server and client parts, wherein the client part resides on one or more user devices (e.g., devices 104, 122, 200, 400, 600, 800, 1000, 1100, or 1200) and communicates with the server part (e.g., server system 108) via one or more networks, for example, as... Figure 1 As shown in the image. In some examples, the digital assistant system 700 is... Figure 1 The specific implementation of server system 108 (and / or DA server 106) shown is illustrated. It should be noted that digital assistant system 700 is only one example of a digital assistant system, and digital assistant system 700 may have more or fewer components than shown, may combine two or more components, or may have different configurations or layouts of components. Figure 7A The various components shown are implemented in hardware, software instructions for execution by one or more processors, firmware (including one or more signal processing integrated circuits and / or application-specific integrated circuits), or a combination thereof.

[0167] The digital assistant system 700 includes a memory 702, an input / output (I / O) interface 706, a network communication interface 708, and one or more processors 704. These components can communicate with each other via one or more communication buses or signal lines 710.

[0168] In some examples, memory 702 includes non-transitory computer-readable media, such as high-speed random access memory and / or non-volatile computer-readable storage media (e.g., one or more disk storage devices, flash memory devices or other non-volatile solid-state storage devices).

[0169] In some examples, I / O interface 706 couples input / output devices 716 of digital assistant system 700 (such as a display, keyboard, touchscreen, and microphone) to user interface module 722. I / O interface 706, in conjunction with user interface module 722, receives user input (e.g., voice input, keyboard input, touch input, etc.) and processes it accordingly. In some examples, such as when the digital assistant is implemented on a standalone user device, digital assistant system 700 includes devices 200, 400, 600, 800, and / or 1000 (…). Figure 2A , Figure 4 , Figures 6A to 6B , Figure 8 and Figures 10A to 10B The components and I / O communication interfaces described herein. In some examples, the digital assistant system 700 represents the server portion of a digital assistant implementation and can interact with the user through a client-side portion residing on a user device (e.g., device 104, 200, 400, 600, 800, 1000, 1100, or 1200).

[0170] In some examples, the network communication interface 708 includes a wired communication port 712 and / or wireless transmitting and receiving circuitry 714. The wired communication port receives and transmits communication signals via one or more wired interfaces such as Ethernet, Universal Serial Bus (USB), FireWire, etc. The wireless circuitry 714 receives RF signals and / or optical signals from the communication network and other communication devices, and transmits RF signals and / or optical signals to the communication network and other communication devices. Wireless communication uses any of a variety of communication standards, protocols, and technologies, such as GSM, EDGE, CDMA, TDMA, Bluetooth, Wi-Fi, VoIP, Wi-MAX, or any other suitable communication protocol. The network communication interface 708 enables the digital assistant system 700 to communicate with other devices via networks such as the Internet, intranets, and / or wireless networks such as cellular telephone networks, wireless local area networks (LANs), and / or metropolitan area networks (MANs).

[0171] In some examples, memory 702 or its computer-readable storage medium stores programs, modules, instructions, and data structures, including all or a subset of the following: operating system 718, communication module 720, user interface module 722, one or more applications 724, and digital assistant module 726. Specifically, memory 702 or its computer-readable storage medium stores instructions for performing the processes described below. One or more processors 704 execute these programs, modules, and instructions, and read from / write to data structures.

[0172] Operating systems 718 (e.g., Darwin, RTXC, LINUX, UNIX, iOS, OS X, WINDOWS, or embedded operating systems such as VxWorks) include various software components and / or drivers for controlling and managing general system tasks (e.g., memory management, storage device control, power management, etc.) and facilitate communication between various hardware, firmware, and software components.

[0173] The communication module 720 facilitates communication between the digital assistant system 700 and other devices via a network communication interface 708. For example, the communication module 720 communicates with electronic devices (such as those in…) Figure 2A , Figure 4 and Figures 6A to 6B The device 200, 400, or 600 shown communicates with the RF circuit 208. The communication module 720 also includes various components for processing data received by the wireless circuit 714 and / or the wired communication port 712.

[0174] The user interface module 722 receives commands and / or input from the user (e.g., from a keyboard, touchscreen, pointing device, controller, and / or microphone) via the I / O interface 706 and generates user interface objects on the display. The user interface module 722 also prepares output (e.g., speech, sound, animation, text, icons, vibration, haptic feedback, lighting, etc.) and delivers it to the user via the I / O interface 706 (e.g., through a display, audio channel, speaker, touchpad, etc.).

[0175] Application 724 includes programs and / or modules configured to be executed by one or more processors 704. For example, if the digital assistant system is implemented on a standalone user device, application 724 includes user applications such as games, calendar applications, navigation applications, or email applications. If the digital assistant system 700 is implemented on a server, application 724 includes, for example, resource management applications, diagnostic applications, or scheduling applications.

[0176] The memory 702 also stores the digital assistant module 726 (or the server portion of the digital assistant). In some examples, the digital assistant module 726 includes the following submodules or subsets or supersets thereof: input / output processing module 728, speech-to-text (STT) processing module 730, natural language processing module 732, dialogue flow processing module 734, task flow processing module 736, service processing module 738, and speech synthesis processing module 740. Each of these modules has access to one or more, or subsets or supersets of the following systems or data and models of the digital assistant module 726: knowledge ontology 760, vocabulary index 744, user data 748, task flow model 754, service model 756, and ASR system 758.

[0177] In some examples, using the processing modules, data, and models implemented in the digital assistant module 726, the digital assistant can perform at least some of the following: converting verbal input into text; identifying user intent expressed in natural language input received from the user; proactively extracting and obtaining information necessary to fully infer the user intent (e.g., by disambiguating words, games, intents, etc.); determining a task flow to satisfy the inferred intent; and executing the task flow to satisfy the inferred intent.

[0178] In some examples, such as Figure 7B As shown, the I / O processing module 728 can... Figure 7A The I / O device 716 in the middle interacts with the user or through Figure 7A The network communication interface 708 interacts with user equipment (e.g., device 104, device 200, device 400, or device 600) to obtain user input (e.g., speech input) and provide a response to the user input (e.g., as speech output). The I / O processing module 728 optionally obtains contextual information associated with the user input from the user equipment along with or shortly after receiving the user input. Contextual information includes user-specific data, vocabulary, and / or preferences related to the user input. In some examples, contextual information also includes the software and hardware states of the user equipment at the time the user request is received, and / or information related to the user's surrounding environment at the time the user request is received. In some examples, the I / O processing module 728 also sends follow-up questions related to the user request to the user and receives answers from the user. When a user request is received by the I / O processing module 728 and the user request includes speech input, the I / O processing module 728 forwards the speech input to the STT processing module 730 (or speech recognizer) for speech-to-text conversion.

[0179] STT processing module 730 includes one or more ASR systems 758. The one or more ASR systems 758 can process speech input received through I / O processing module 728 to produce recognition results. Each ASR system 758 includes a front-end speech preprocessor. The front-end speech preprocessor extracts representative features from the speech input. For example, the front-end speech preprocessor performs a Fourier transform on the speech input to extract spectral features characterizing the speech input as a sequence of representative multidimensional vectors. Furthermore, each ASR system 758 includes one or more speech recognition models (e.g., acoustic models and / or language models) and implements one or more speech recognition engines. Examples of speech recognition models include Hidden Markov Models, Gaussian Mixture Models, Deep Neural Network Models, n-gram grammar language models, and other statistical models. Examples of speech recognition engines include engines based on Dynamic Time Warping (VTW) and engines based on Weighted Finite State Transformers (WFST). One or more speech recognition models and one or more speech recognition engines are used to process the representative features extracted by the front-end speech preprocessor to produce intermediate recognition results (e.g., phonemes, phoneme strings, and sub-words) and ultimately produce text recognition results (e.g., words, word strings, or token sequences). In some examples, the speech input is processed at least in part by a third-party service or on the user's device (e.g., device 104, device 200, device 400, or device 600) to produce the recognition results. Once the STT processing module 730 produces the recognition results containing text strings (e.g., words, or word sequences, or token sequences), the recognition results are passed to the natural language processing module 732 for intent inference. In some examples, the STT processing module 730 produces multiple candidate text representations of the speech input. Each candidate text representation is a sequence of words or tokens corresponding to the speech input. In some examples, each candidate text representation is associated with a speech recognition confidence score. Based on the speech recognition confidence score, the STT processing module 730 ranks the candidate text representations and provides the n best (e.g., the n highest-ranked) candidate text representations to the natural language processing module 732 for intent inference, where n is a predetermined integer greater than zero. For example, in one example, only the highest-ranked (n=1) candidate text representation is delivered to the natural language processing module 732 for intent inference. In another example, the five highest-ranked (n=5) candidate text representations are passed to the natural language processing module 732 for intent inference.

[0180] Further details regarding speech-to-text processing are described in U.S. Utility Model Patent Application Serial No. 13 / 236,942, entitled "Consolidating Speech Recognition Results," filed September 20, 2011, the entire disclosure of which is incorporated herein by reference.

[0181] In some examples, the STT processing module 730 includes a vocabulary of recognizable words and / or accesses that vocabulary via a phonetic alphabet conversion module 731. Each vocabulary word is associated with one or more candidate pronunciations of a word represented in a speech recognition phonetic alphabet. Specifically, the vocabulary of recognizable words includes words associated with multiple candidate pronunciations. For example, the vocabulary includes words associated with / t 'me o / and / t 'm to The candidate pronunciation of / is associated with the word "tomato". Additionally, lexical words are associated with custom candidate pronunciations based on previous speech input from the user. These custom candidate pronunciations are stored in the STT processing module 730 and associated with a specific user via a user profile on the device. In some examples, candidate pronunciations are determined based on the spelling of the word and one or more linguistic and / or phonetic rules. In some examples, candidate pronunciations are manually generated, for example, based on known canonical pronunciations.

[0182] In some examples, candidate pronunciations are ranked based on their commonness. For example, the candidate pronunciation / t 'me o / Ranked higher than / t 'm to / is chosen because the former is (e.g., among all users, for users in a specific geographic region, or for any other suitable subset of users) a more common pronunciation. In some examples, candidate pronunciations are ranked based on whether they are custom candidate pronunciations associated with a user. For example, custom candidate pronunciations rank higher than canonical candidate pronunciations. This can be useful for identifying proper nouns with unique pronunciations that deviate from the canonical pronunciation. In some examples, candidate pronunciations are associated with one or more speech characteristics such as geographic origin, country, or ethnicity. For example, the candidate pronunciation / t 'me o / Related to the United States, while candidate pronunciation / t 'm to / is associated with the United Kingdom. Furthermore, the ranking of candidate pronunciations is based on one or more characteristics of the user (e.g., geographic origin, country, ethnicity, etc.) stored in the user profile on the device. For example, a user might be associated with the United States based on their user profile. Based on the user's association with the United States, the candidate pronunciation / t 'me o / (related to the United States) ranked higher than the candidate pronunciation / t 'm to / (associated with the UK). In some examples, one of the ranked candidate pronunciations is selected as the predicted pronunciation (e.g., the most likely pronunciation).

[0183] Upon receiving speech input, the STT processing module 730 is used (e.g., using an acoustic model) to determine the phonemes corresponding to the speech input, and then attempts (e.g., using a language model) to determine the words that match those phonemes. For example, if the STT processing module 730 first identifies a phoneme sequence / t corresponding to a portion of the speech input... 'me o Then it can then determine, based on the vocabulary index 744, that the sequence corresponds to the word "tomato".

[0184] In some examples, the STT processing module 730 uses approximate matching techniques to determine words in a utterance. Thus, for example, the STT processing module 730 determines the phoneme sequence / t 'me o / corresponds to the word "tomato", even if the specific phoneme sequence is not one of the candidate phoneme sequences for the word.

[0185] The digital assistant's natural language processing module 732 ("natural language processor") acquires n best candidate text representations ("word sequences" or "lexical sequences") generated by the STT processing module 730 and attempts to associate each candidate text representation with one or more "executable intents" identified by the digital assistant. An "executable intent" (or "user intent") represents a task that can be performed by the digital assistant and may have an associated task flow implemented in the task flow model 754. An associated task flow is a series of programmed actions and steps taken by the digital assistant to perform the task. The capabilities of the digital assistant depend on the number and variety of task flows implemented and stored in the task flow model 754, or in other words, on the number and variety of "executable intents" identified by the digital assistant. However, the effectiveness of the digital assistant also depends on its ability to infer the correct "executable intent" from a user request expressed in natural language.

[0186] In some examples, in addition to the word or lexical sequences obtained from the STT processing module 730, the natural language processing module 732 also receives, for example, contextual information associated with the user request from the I / O processing module 728. The natural language processing module 732 may optionally use the contextual information to clarify, supplement, and / or further define the information contained in the candidate text representation received from the STT processing module 730. Contextual information includes, for example, user preferences, the hardware and / or software state of the user's device, sensor information collected before, during, or shortly after the user request, previous interactions (e.g., conversations) between the digital assistant and the user, etc. As described herein, in some examples, the contextual information is dynamic and varies with the time, location, content, and other factors of the conversation.

[0187] In some examples, natural language processing is based on, for example, a knowledge ontology 760. Knowledge ontology 760 is a hierarchical structure containing many nodes, each node representing an "executable intent" or an "attribute" related to one or more of the "executable intent" or other "attributes." As noted above, an "executable intent" represents a task that a digital assistant can perform; that is, the task is "executable" or can be done. An "attribute" represents a parameter associated with a sub-aspect of an executable intent or another attribute. The links between executable intent nodes and attribute nodes in knowledge ontology 760 define how the parameters represented by the attribute nodes are related to the task represented by the executable intent nodes.

[0188] In some examples, the knowledge ontology 760 consists of executable intent nodes and attribute nodes. Within the knowledge ontology 760, each executable intent node is linked directly or through one or more intermediate attribute nodes to one or more attribute nodes. Similarly, each attribute node is linked directly or through one or more intermediate attribute nodes to one or more executable intent nodes. For example, as... Figure 7C As shown, knowledge ontology 760 includes a "Restaurant Reservation" node (i.e., an executable intent node). The attribute nodes "Restaurant", "Date / Time" (for reservations) and "Participant Size" are each directly linked to the executable intent node (i.e., the "Restaurant Reservation" node).

[0189] Furthermore, the attribute nodes "Cuisine," "Price Range," "Phone Number," and "Location" are child nodes of the attribute node "Restaurant," and all are linked to the "Restaurant Reservation" node (i.e., the executable intent node) through the intermediate attribute node "Restaurant." For example, ... Figure 7CAs shown, knowledge ontology 760 also includes a "Set Reminder" node (i.e., another executable intent node). The attribute nodes "Date / Time" (for setting a reminder) and "Topic" (for the reminder) are each linked to the "Set Reminder" node. Since the attribute "Date / Time" is related to both the task of making a restaurant reservation and the task of setting a reminder, the attribute node "Date / Time" is linked to both the "Restaurant Reservation" node and the "Set Reminder" node in knowledge ontology 760.

[0190] An executable intent node, along with its linked attribute nodes, is described as a "domain." In this discussion, each domain is associated with a corresponding executable intent and refers to a set of nodes (and the relationships between these nodes) associated with a particular executable intent. For example, Figure 7C The knowledge ontology 760 shown includes examples of a restaurant reservation domain 762 and a reminder domain 764 within the knowledge ontology 760. The restaurant reservation domain includes an actionable intent node “Restaurant Reservation”, attribute nodes “Restaurant”, “Date / Time”, and “Participant Size”, and sub-attribute nodes “Cuisine”, “Price Range”, “Phone Number”, and “Location”. The reminder domain 764 includes an actionable intent node “Set Reminder” and attribute nodes “Topic” and “Date / Time”. In some examples, the knowledge ontology 760 consists of multiple domains. Each domain shares one or more attribute nodes with one or more other domains. For example, in addition to the restaurant reservation domain 762 and the reminder domain 764, the “Date / Time” attribute node is associated with many different domains (e.g., scheduling domain, travel booking domain, movie ticket domain, etc.).

[0191] although Figure 7C Two example fields within knowledge ontology 760 are illustrated, but other fields include, for example, "Find a movie," "Initiate a phone call," "Find directions," "Schedule a meeting," "Send a message," and "Provide answers to questions," "Reading lists," "Provide navigation instructions," and "Provide instructions for a task," etc. The "Send a message" field is associated with the "Send a message" executable intent node and also includes attribute nodes such as "Recipient," "Message type," and "Message body." The attribute node "Recipient" is further defined, for example, by sub-attribute nodes such as "Recipient name" and "Message address."

[0192] In some examples, knowledge ontology 760 includes all domains (and therefore executable intents) that the digital assistant can understand and manipulate. In some examples, knowledge ontology 760 is modified, such as by adding or removing entire domains or nodes, or by modifying the relationships between nodes within knowledge ontology 760.

[0193] In some examples, nodes associated with multiple related executable intents are clustered under a “superdomain” in Knowledge Ontology 760. For example, the “Travel” superdomain includes clusters of travel-related attribute nodes and executable intent nodes. Travel-related executable intent nodes include “flight booking,” “hotel booking,” “car rental,” “route planning,” “finding points of interest,” and so on. Executable intent nodes under the same superdomain (e.g., the “Travel” superdomain) share multiple attribute nodes. For example, executable intent nodes for “flight booking,” “hotel booking,” “car rental,” “get route,” and “finding points of interest” share one or more of the attribute nodes “starting location,” “destination,” “departure date / time,” “arrival date / time,” and “party size.”

[0194] In some examples, each node in the knowledge ontology 760 is associated with a set of words and / or phrases related to the attribute or executable intent represented by the node. The corresponding set of words and / or phrases associated with each node is called the "vocabulary" associated with the node. The corresponding set of words and / or phrases associated with each node is stored in the vocabulary index 744 associated with the attribute or executable intent represented by the node. For example, returning... Figure 7B The vocabulary associated with the node for the attribute "restaurant" includes words such as "food," "drinks," "cuisine," "hunger," "eat," "pizza," "fast food," and "meal." Similarly, the vocabulary associated with the node for the executable intent "initiate a phone call" includes words and phrases such as "call," "make a phone call," "dial," "talk to," "call this number," and "make a phone call." The vocabulary index 744 optionally includes words and phrases from different languages.

[0195] Natural Language Processing (NLP) module 732 receives candidate text representations (e.g., text strings or lexical sequences) from STT processing module 730 and, for each candidate representation, determines which nodes the words in the candidate text representation relate to. In some examples, if a word or phrase in the candidate text representation is found to be associated with one or more nodes in knowledge ontology 760 (via lexical index 744), the word or phrase "triggers" or "activates" those nodes. Based on the number and / or relative importance of the activated nodes, NLP module 732 selects one executable intent as the task the user intends the digital assistant to perform. In some examples, the domain with the most "triggered" nodes is selected. In some examples, the domain with the highest confidence (e.g., based on the relative importance of its individual triggered nodes) is selected. In some examples, the domain is selected based on a combination of the number and importance of the triggered nodes. In some examples, additional factors, such as whether the digital assistant has previously correctly interpreted similar requests from the user, are also considered in the node selection process.

[0196] User data 748 includes user-specific information such as user-specific vocabulary, user preferences, user address, user's default and auxiliary languages, user's contact list, and other short-term or long-term information for each user. In some examples, the natural language processing module 732 uses user-specific information to supplement the information contained in the user input to further define the user's intent. For example, for a user request "Invite my friends to my birthday party," the natural language processing module 732 can access user data 748 to determine who the "friends" are and when and where the "birthday party" will be held, rather than requiring the user to explicitly provide such information in his / her request.

[0197] It should be recognized that, in some examples, the natural language processing module 732 is implemented using one or more machine learning mechanisms (e.g., neural networks). Specifically, the one or more machine learning mechanisms are configured to receive candidate text representations and contextual information associated with those candidate text representations. Based on the candidate text representations and the associated contextual information, the one or more machine learning mechanisms are configured to determine an intent confidence score for a set of candidate executable intents. The natural language processing module 732 may select one or more candidate executable intents from this set of candidate executable intents based on the determined intent confidence score. In some examples, a knowledge ontology (e.g., knowledge ontology 760) is also used to select one or more candidate executable intents from this set of candidate executable intents.

[0198] Further details regarding the search of knowledge ontology based on word strings are described in U.S. Utility Model Patent Application Serial No. 12 / 341743, entitled “Method and Apparatus for Searching Using An Active Ontology,” filed on December 22, 2008, the entire disclosure of which is incorporated herein by reference.

[0199] In some examples, once the natural language processing module 732 identifies an executable intent (or domain) based on a user request, it generates a structured query to represent the identified executable intent. In some examples, the structured query includes parameters for one or more nodes within the domain of the executable intent, and at least some of these parameters are populated with specific information and requirements specified in the user request. For example, a user says, "Make a dinner reservation for me at a sushi restaurant at 7 pm." In this case, the natural language processing module 732 is able to correctly identify the executable intent as "restaurant reservation" based on the user input. According to the knowledge ontology, the structured query for the "restaurant reservation" domain includes parameters such as {cuisine}, {time}, {date}, {number of people}, etc. In some examples, based on verbal input and text derived from the verbal input using the STT processing module 730, the natural language processing module 732 generates a partially structured query for the restaurant reservation domain, where the partially structured query includes the parameters {cuisine = "sushi"} and {time = "7 pm"}. However, in this example, the user's utterance contains insufficient information to complete a structured query associated with the domain. Therefore, based on the currently available information, no other necessary parameters, such as {number of people at the party} and {date}, are specified in the structured query. In some examples, the natural language processing module 732 uses the received context information to populate some parameters of the structured query. For example, in some examples, if a user requests "sushi restaurants near me," the natural language processing module 732 populates the {location} parameter in the structured query with GPS coordinates from the user's device.

[0200] In some examples, the Natural Language Processing (NLP) module 732 identifies multiple candidate executable intents for each candidate text representation received from the STT processing module 730. Furthermore, in some examples, a corresponding structured query (partially or entirely) is generated for each identified candidate executable intent. The NLP module 732 determines an intent confidence score for each candidate executable intent and ranks the candidate executable intents based on the intent confidence scores. In some examples, the NLP module 732 passes one or more of the generated structured queries (including any completed parameters) to the task flow processing module 736 (“task flow processor”). In some examples, one or more structured queries for the m best (e.g., the m highest-ranked) candidate executable intents are provided to the task flow processing module 736, where m is a predetermined integer greater than zero. In some examples, one or more structured queries for the m best candidate executable intents, along with their corresponding candidate text representations, are provided to the task flow processing module 736.

[0201] Further details regarding the inference of user intent based on multiple candidate executable intents determined from multiple candidate text representations of speech input are described in U.S. Utility Model Patent Application Serial No. 14 / 298,725, filed June 6, 2014, entitled “System and Method for Inferring UserIntent From Speech Inputs,” the entire disclosure of which is incorporated herein by reference.

[0202] Task flow processing module 736 is configured to receive one or more structured queries from natural language processing module 732, complete the structured queries if necessary, and perform the actions required to "complete" the user's final request. In some examples, various processes required to complete these tasks are provided in task flow model 754. In some examples, task flow model 754 includes processes for obtaining additional information from the user and task flows for performing actions associated with executable intentions.

[0203] As described above, to complete a structured query, task flow processing module 736 needs to initiate additional dialogue with the user to obtain additional information and / or clarify potentially ambiguous statements. When such interaction is required, task flow processing module 736 invokes dialogue flow processing module 734 to participate in the dialogue with the user. In some examples, dialogue flow processing module 734 determines how (and / or when) to request additional information from the user and receives and processes user responses. I / O processing module 728 presents questions to the user and receives answers from the user. In some examples, dialogue flow processing module 734 presents dialogue output to the user via audible and / or visual output and receives input from the user via spoken or physical (e.g., click) responses. Continuing with the above examples, when task flow processing module 736 invokes dialogue flow processing module 734 to determine the "party size" and "date" information for a structured query associated with the domain "restaurant reservations," dialogue flow processing module 734 generates questions such as "How many people?" and "On which day?" to deliver to the user. Once a response is received from the user, the dialogue flow processing module 734 fills the missing information into the structured query, or passes the information to the task flow processing module 736 to complete the missing information in the structured query.

[0204] Once the task flow processing module 736 has completed the structured query for the executable intent, it continues to execute the final task associated with the executable intent. Therefore, the task flow processing module 736 executes the steps and instructions in the task flow model based on the specific parameters contained in the structured query. For example, the task flow model for the executable intent "restaurant reservation" includes steps and instructions for contacting the restaurant and actually requesting a reservation for a specific number of people at a specific party at a specific time. For example, using a structured query such as: {restaurant reservation, restaurant = ABC Cafe, date = 3 / 12 / 2012, time = 7 pm, number of people = 5}, the task flow processing module 736 can perform the following steps: (1) log in to ABC Cafe's server or such as OPENTABLE ® The restaurant reservation system (2) enters the date, time and party number information on the website, (3) submits the form, and (4) creates a calendar entry for the reservation in the user's calendar.

[0205] In some examples, the task flow processing module 736 employs the assistance of the service processing module 738 (“Service Processing Module”) to complete a task requested in the user input or to provide an informative response to a request in the user input. For example, the Service Processing Module 738 may initiate a phone call on behalf of the Task Flow Processing Module 736, set a calendar entry, invoke a map search, invoke or interact with other user applications installed on the user's device, or invoke or interact with third-party services (e.g., restaurant reservation portals, social networking sites, bank portals, etc.). In some examples, the protocols and application programming interfaces (APIs) required for each service are specified through the corresponding service model in Service Model 756. The Service Processing Module 738 accesses the appropriate service model for the service and, based on the service model, generates a request for the service according to the protocols and APIs required by the service.

[0206] For example, if a restaurant has enabled an online reservation service, it submits a service model specifying the necessary parameters for making a reservation and the values ​​of those parameters to be transmitted to the online reservation service's API. When requested by the task flow processing module 736, the service processing module 738 can use the web address stored in the service model to establish a network connection with the online reservation service and send the necessary reservation parameters (e.g., time, date, number of party members) to the online reservation interface in a format appropriate to the online reservation service's API.

[0207] In some examples, the natural language processing module 732, the dialogue flow processing module 734, and the task flow processing module 736 are used jointly and iteratively to infer and define user intent, obtain information to further clarify and refine the user intent, and ultimately generate a response (i.e., output to the user or completion of a task) to satisfy the user intent. The generated response is a dialogue response to the verbal input that at least partially satisfies the user intent. Furthermore, in some examples, the generated response is output as verbal output. In these examples, the generated response is sent to the speech synthesis processing module 740 (e.g., a speech synthesizer), which processes the generated response to synthesize the dialogue response in verbal form. In still other examples, the generated response is data content related to satisfying a user request in the verbal input.

[0208] In an example where task flow processing module 736 receives multiple structured queries from natural language processing module 732, task flow processing module 736 initially processes a first structured query among the received structured queries to attempt to complete the first structured query and / or execute one or more tasks or actions represented by the first structured query. In some examples, the first structured query corresponds to the highest-ranking executable intent. In other examples, the first structured query is selected from structured queries received based on a combination of a corresponding speech recognition confidence score and a corresponding intent confidence score. In some examples, if task flow processing module 736 encounters an error during processing of the first structured query (e.g., due to the inability to determine necessary parameters), task flow processing module 736 may continue to select and process a second structured query among the received structured queries that corresponds to a lower-ranking executable intent. For example, the second structured query may be selected based on a speech recognition confidence score of a corresponding candidate text representation, an intent confidence score of a corresponding candidate executable intent, missing necessary parameters in the first structured query, or any combination thereof.

[0209] Speech synthesis processing module 740 is configured to synthesize speech output for presentation to a user. Speech synthesis processing module 740 synthesizes speech output based on text provided by a digital assistant. For example, the generated dialogue response is in the form of a text string. Speech synthesis processing module 740 converts the text string into audible speech output. Speech synthesis processing module 740 uses any appropriate speech synthesis techniques to generate speech output from text, including but not limited to: concatenation synthesis, unit selection synthesis, diphone synthesis, domain-specific synthesis, formant synthesis, articulation synthesis, Hidden Markov Model (HMM) based synthesis, and sine wave synthesis. In some examples, speech synthesis processing module 740 is configured to synthesize individual words based on phoneme strings corresponding to those words. For example, a phoneme string is associated with a word in the generated dialogue response. The phoneme string is stored in metadata associated with that word. Speech synthesis processing module 740 is configured to directly process the phoneme string in the metadata to synthesize words in speech form.

[0210] In some examples, as an alternative to (or as a supplement to) using the speech synthesis processing module 740, speech synthesis is performed on a remote device (e.g., server system 108), and the synthesized speech is sent to a user device for output to the user. This could occur, for example, in some implementations where output for a digital assistant is generated at the server system. Moreover, since the server system typically has greater processing power or more resources than the user device, it is possible to achieve higher quality speech output than client-side synthesis would achieve.

[0211] Additional details regarding digital assistants can be found in U.S. Utility Model Patent Application No. 12 / 987,982, filed January 10, 2011, entitled “Intelligent Automated Assistant,” and U.S. Utility Model Patent Application No. 13 / 251,088, filed September 30, 2011, the entire disclosure of which is incorporated herein by reference.

[0212] 4. Contextual speech interpretation using large language models

[0213] Figure 8Computer system 800 is illustrated according to various examples. Computer system 800 includes at least some components of devices 104, 122, 200, 400, and / or 600. In some examples, computer system 800 is implemented as a smartphone, tablet, laptop, desktop computer, smartwatch, smart home device (e.g., smart speaker, smart TV, and / or smart home appliance), wearable device, head-mounted device, or another type of mobile device (e.g., car, robot, and / or drone). Although Figure 8 Some components of computer system 800 are illustrated, but those skilled in the art will understand that some components of computer system 800 are not illustrated. For the sake of brevity and to avoid obscuring more relevant aspects of this disclosure, such components are not illustrated.

[0214] also, Figure 8 This includes functional descriptions of various components that may exist in a particular example, which differ from the structural diagrams of the examples described herein. As those skilled in the art will recognize, the components shown individually can be combined, and some components can be separated. For example, Figure 8 Some functional components shown individually can be implemented in a single component, and the various functions of a single functional component can be implemented through one or more functional components in various examples. The actual number of components and how functions are distributed among the components will vary depending on the specific implementation, and in some examples, depends in part on the specific combination of hardware, software, and / or firmware chosen for a particular specific implementation of the computer system 800.

[0215] Computer system 800 at least partially implements digital assistant system 700, as mentioned above. Figures 7A to 7C As described herein, computer system 800 also includes memory 802. Memory 802 is at least partially implemented as a separate instance of at least one of memories 202, 470, 618, and 702. For example, memory 802 includes at least some components of memories 202, 470, 618, and / or 702. In some examples, memory 802 includes one or more additional components described below that enable computer system 800 to incorporate large language models into a digital assistant system and perform the processes described herein.

[0216] In a non-limiting example, memory 802 includes an eye-tracking unit 804, and computer system 800 includes an eye-tracking device 806. Eye-tracking unit 804 includes instructions and / or logic for tracking the positioning and movement of a user's gaze (or more broadly, the user's eyes, face, or head) using data acquired from eye-tracking device 806. In some examples, eye-tracking unit 804 tracks the positioning and movement of the user's gaze relative to the physical environment, relative to the user (e.g., the user's hands, face, or head), relative to computer system 800, and / or relative to content displayed by display generation components.

[0217] Eye-tracking device 806 is controlled by eye-tracking unit 804 and includes various hardware and / or software components configured to perform eye-tracking technology. For example, eye-tracking device 806 includes at least one eye-tracking camera (e.g., an infrared (IR) or near-infrared (NIR) camera) and an illumination source (e.g., an array or ring of IR or NIR light sources, such as LEDs) that emits light (e.g., IR or NIR light) toward the user's eye. The eye-tracking camera may be pointed at the user's eye to receive IR or NIR light reflected directly from the eye, or alternatively, it may be pointed at a mirror that reflects IR or NIR light from the eye back to the eye-tracking camera. Eye-tracking device 806 may optionally capture images of the user's eyes (e.g., as a video stream captured at 60 to 120 frames per second), analyze these images to generate eye-tracking information, and transmit the eye-tracking information to eye-tracking unit 804. In some examples, the user's two eyes are tracked separately using corresponding eye-tracking cameras and illumination sources. In some examples, only one of the user's eyes is tracked using corresponding eye-tracking cameras and illumination sources.

[0218] In a non-limiting example, memory 802 includes a hand tracking unit 808, and computer system 800 includes a hand tracking device 810. Hand tracking unit 808 includes instructions and / or logic for tracking the positioning and / or movement of one or more parts of a user's hand using hand tracking data acquired from hand tracking device 810. Hand tracking unit 808 tracks positioning and / or movement relative to the physical environment, relative to the user (e.g., the user's head, face, or eyes), relative to computer system 800, relative to content displayed by display generation components, and / or relative to a coordinate system defined about the user's hand. In some examples, hand tracking unit 808 analyzes hand tracking data to identify hand gestures (e.g., aerial gestures, such as grasping, pinching, or pointing gestures) and identifies content corresponding to the hand gestures (e.g., physical or virtual content), such as content selected by the hand gesture.

[0219] The hand tracking device 810 is controlled by the hand tracking unit 808 and includes various hardware and / or software components configured to perform hand tracking and hand pose recognition technologies. For example, the hand tracking device 810 includes one or more image sensors (e.g., one or more IR cameras, 3D cameras, depth cameras, and / or color cameras, etc.) that capture three-dimensional information (e.g., depth maps) representing the human user's hand. The one or more image sensors capture hand images at sufficient resolution to distinguish fingers and their corresponding positions. In some examples, the one or more image sensors project a blob pattern onto the environment including the hand and capture images of the projected pattern. In some examples, the one or more image sensors capture time series of hand tracking data (e.g., the captured three-dimensional information and / or the captured images of the projected pattern), and the hand tracking device 810 transmits the time series of hand tracking data to the hand tracking unit 808 for further analysis, such as to identify hand poses, hand postures, and / or hand movements.

[0220] In some examples, the hand tracking device 810 includes one or more hardware input devices configured to be worn and / or held (or otherwise attached to one or more corresponding hands of a user). In such examples, the hand tracking device 810 tracks the position, posture, and / or movement of the user's hand based on tracking the position, posture, and / or movement of the corresponding hardware input device. The hand tracking device 810 tracks the position, posture, and / or movement of the corresponding hardware input device optically (e.g., via one or more image sensors) and / or based on data obtained from sensors (e.g., accelerometers, magnetometers, gyroscopes, inertial measurement units, etc.) contained within the hardware input device. In some examples, the hardware input device includes one or more physical controls (e.g., buttons, touch-sensitive surfaces, pressure-sensitive surfaces, knobs, joysticks, etc.), and user input selecting a particular physical control provides functionality similar to input of the corresponding hand posture. For example, if computer system 800 interprets a pinched hand gesture input as a user selection of a specific element, computer system 800 may alternatively or additionally interpret a selection of a physical button on a hardware device as a user selection of a specific element.

[0221] Although Figure 8 As illustrated, the DA system 700, memory 802 (including eye-tracking unit 804, head-tracking unit 808, and LLM unit 812), eye-tracking device 806, and head-tracking device 810 reside on a single computing device; however, in other examples, the components and / or functionality of the computer system 800 may be distributed across multiple computing devices in various different ways. For example, server system 108 ( Figure 1The LLM unit 812 and its associated functionality are implemented in one device, while other components of the computer system 800 reside on a separate computing device that is separate from and communicates with the server system 108. As another example, eye-tracking device 806 and hand-tracking device 810 reside on a first device that communicates with a separate second device that implements the other components of the computer system 800 and their associated functions.

[0222] Figure 9 An exemplary base system 900 including a base model 910 is illustrated according to some embodiments. In some embodiments, blocks of the base system 900 are combined, the order of the blocks is changed, and / or blocks of the base system 900 are removed.

[0223] The base system 900 includes a lexicalization module 906 that uses input data 902, an input embedding module 908, and a base model 910, and optionally includes a context module 904 for training the base model 910 to process the input data 902 to determine the output 912.

[0224] In some examples, various components of the digital assistant system 700 include generative artificial intelligence (AI) (such as a base model 910) and / or utilize that generative AI. In some examples, the base model 910 includes a subset of machine learning models trained to generate text, images, and / or other media based on a training dataset comprising a large amount of specific types of data. The base model 910 is then integrated into (or otherwise made available to) components of the digital assistant system 700 to provide the digital assistant system 700 with text, images, and / or other media as output for determining, performing, and / or providing tasks.

[0225] Generative AI models (such as the base model 910) utilize self-supervised or semi-supervised learning to be trained on large amounts of data to adapt to specific downstream tasks. For example, the base model 910 is trained using a large set of different images and corresponding text or metadata to determine a description of newly captured image data as output 912. The digital assistant system 700 can then use these descriptions to determine user intent, tasks, and / or other information that can be used to perform the tasks.

[0226] The base model is typically trained first using a large unlabeled dataset and then later adapted to specific tasks within the architecture of the digital assistant system 700. Therefore, specific tasks or output types are not encoded into the base model; instead, the trained base model emerges from self-supervised training using unlabeled data. The trained base model is then adapted to various tasks based on the needs of the digital assistant system 700 to efficiently perform tasks for the user.

[0227] Large Language Models (LLMs) are a type of foundational model that provides text output after being trained on a large set of input text data. Like other foundational models, LLMs can be trained in a self-supervised manner, and therefore different LLMs trained on the same large set of input text can produce different outputs. These LLMs can then be adapted for use with a digital assistant system 700 to suit specific types of text. Thus, in some examples, the LLM is trained to determine a summary of the text provided as input, while in other examples, the LLM is trained to predict text based on the set of input text. Therefore, LLMs can efficiently process large amounts of input text to provide the digital assistant with text that can be used to determine and / or perform tasks.

[0228] In some examples, the LLM can be trained in a semi-supervised manner and / or provided with human feedback to refine its output. This allows the LLM to be adapted to provide specific outputs required for a particular task of the digital assistant system 700, such as summarizing large amounts of text or tasks that the digital assistant system 700 needs to perform. Furthermore, the inputs provided to the LLM can be adapted so that the LLM processes data as efficiently as, or more efficiently than, the digital assistant system 700 could process data without the LLM.

[0229] Once the base model 910 (e.g., LLM) has been fully trained, the base model 910 can process the input data 902 as discussed below to determine the output 912, which can be used to further train the base model 910 or can be processed by the digital assistant 700 to perform tasks and / or provide output to the user.

[0230] Specifically, input data 902 is received and provided to lexicalization module 906, which converts the input data 902 into lexical units and / or a series of lexical units in a format that can be processed by input embedding module 908 into a format understood by base model 910. Lexicalization module 906 converts the input data into a series of characters that have specific semantic meaning to base model 910.

[0231] In some examples, the lexicalization module 906 lexicalizes the context data from the context module 904 to add additional information to the input data 902 for processing by the base model 910. For example, the context module 904 may provide information related to the input data 902, such as the location where the input data 902 was received, the time when the input data 902 was received, other data received simultaneously with the input data 902, and / or other contextual information related to the input data 902. The lexicalization module 906 can then lexicalize this context data using the input data 902 to provide it to the base model 910.

[0232] After the input data 902 has been lexicalized, it is provided to the input embedding module 908 to convert the lexies into vector representations that can be processed by the base model 910. In some examples, the vector representations include information provided by the context module 904. In some examples, the vector representations include information determined from the output 912. Thus, the input embedding module 908 converts the various data provided as input into a format that the base model 910 can parse and process.

[0233] For example, when the base model 910 is a large language model (LLM), the lexicalization module 906 converts the input data 902 into text, which is then converted into a vector representation by the input embedding module 908. This vector representation can be processed by the base model 910 to determine a response to the input data 902 as output 912, or to determine a summary of the input data 902 as output 912. As another example, when the base model 910 is a model trained to determine a description of an image, the image input data 902 can be lexicalized into characters and then converted into a vector representation by the input embedding module 908. This vector representation is processed by the base model 910 to determine a description of the image as output 912.

[0234] The base model 910 uses a series of layers to process the received vector representation. In some implementations, this series of layers includes an attention layer 910a, a normalization layer 910b, a feedforward layer 910c, and / or a normalization layer 910d. In some examples, the base model 910 additionally includes layers similar to these to further process the vector representation. Therefore, the base model 910 can be customized based on the specific task it has been trained to perform. Each layer in the base model 910 performs a specific task to process the vector representation into output 912.

[0235] Attention layer 910a provides access to all parts of the vector representation simultaneously, thereby increasing the speed at which the vector representation can be processed and ensuring that data is processed equally across all parts of the vector representation. Normalization layers 910b and 910d scale the data processed by the base model 910 up or down based on the needs of the other layers of the base model 910. This allows the base model 910 to manipulate the data as needed during processing. Feedforward layer 910c assigns weights to the data being processed and provides that data for further processing within the base model 910. These layers work together to process the vector representation provided to the base model 910, thereby determining the appropriate output 912.

[0236] For example, as discussed above, when the base model 910 is a large language model (LLM), the base model 910 processes the input text to determine a summary and / or additional follow-up text as output 912. As another example, as discussed above, when the base model 910 is a model trained to determine a description of an image, the base model 910 processes the input image to determine a description of the image and / or a task that can be performed based on the content of the image as output 912.

[0237] In some examples, output 912 is further processed by digital assistant system 700 to provide output or perform tasks. For instance, when output 912 is a statement describing a task that digital assistant system 700 has performed, digital assistant system 700 can use the text to create visual or audio output to be provided to the user. As another example, when output 912 is text including a function and the parameters of that function, digital assistant system 700 can execute a function call to perform the function with the provided parameters.

[0238] In some examples, the digital assistant system 700 includes multiple generative AI (e.g., foundational) models that work together to process data efficiently. In some examples, a component of the digital assistant system 700 can be replaced by a generative AI (e.g., foundational) model trained to perform the same function as that component. In some examples, these generative AI models are more efficient than traditional components and / or provide the digital assistant system 700 with more flexible processing and / or outputs to utilize.

[0239] 5. Contextual speech interpretation using large language models

[0240] Figures 10A to 10B Techniques for contextual speech interpretation using large language models are illustrated according to various examples. For example, device 1002 may include any device described herein, including but not limited to devices 104, 200, 400, and 600. Figure 1 , Figure 2A , Figure 4 and Figures 6A to 6B Therefore, it should be understood that, with Figures 10A to 10B The associated device 1002 may correspond to a user device, such as a laptop computer, tablet computer, telephone, etc. Furthermore, the processes described herein may be performed by a server having information delivered to and from the device, information executed on the device, or a combination thereof. In some embodiments, the first and second electronic devices may reside within a single device (e.g., a head-mounted display with a microphone, speakers, etc.).

[0241] refer to Figure 10ASystem 1000 includes a user 1004 that interacts with a user device 1002 (such as a smartphone). In this example, the user communicates with another user via a messaging application. The other user is identified by an identifier within the messaging application, such as identifier 1006 for a contact named "John". When interacting with device 1002, the user provides multimodal input to perform various functions, such as using touch input, gaze input, speech input, device movement, etc. For example, when viewing the messaging application, the user can request a digital assistant to respond to the user associated with identifier 1006 via speech input. Once speech input is received, a representation of the speech input is obtained, such as a text representation processed via STT or other representations that the system can recognize and process.

[0242] In some cases, the representation of verbal input includes various features that can make natural language processing more difficult, such as disfluency, corrections, or other ambiguities. For example, user 1004 provides the verbal input “Hey assistant, call him. I mean, message him.” In this example, the user is providing corrections to the digital assistant indicating that the user does not intend to call another user but intends to send a text message to another user. The verbal input also includes the ambiguous term “he.” Other examples include ambiguous user intent, such as “How’s New York?” Here, the intent is ambiguous because it is unclear what the user wants to know about New York. In this example, the user may have previously provided verbal input including “How’s the weather in San Francisco?” Therefore, ambiguous user intent can be resolved by analyzing the content of the conversation. Other examples include disfluency, such as “Hey assistant, uh, help me find…wait a minute…a good restaurant in downtown Seattle.” Disfluency “uh” and “wait a minute” indicate utterances that cause interruptions or otherwise disrupt the flow of verbal input. In normal human interactions, these disfluency effects may be minor, but in natural language processing (NLP), they can lead to problems in speech interpretation (e.g., an NLP system may be unable to decipher the intent of a speech based on its inability to interpret the meaning of "uh" and "wait a minute").

[0243] Once a representation of the speech input is obtained, it is determined whether entities (such as words or phrases) within the speech representation require disambiguation. For example, the speech representation in this example includes "call him. I mean, message him." The input includes the ambiguous term "he," making disambiguation necessary to determine the contact the user is referring to. Depending on the degree to which entities within the speech representation require disambiguation, a referencing disambiguation system is used to attempt to disambiguate the ambiguity. In this example, based on the concurrent use of the messaging application with an identifier 1006 for a contact named "John," the referencing disambiguation system is able to disambiguate the meaning of the ambiguous term "he" to indicate that the user is referring to the contact associated with identifier 1006. Therefore, the referencing disambiguation system adjusts the speech representation by adding indications to the corresponding entities. Specifically, the referencing disambiguation system may include parentheses or other indications appended to ambiguous terms to indicate that the referencing disambiguation system has disambiguated the terms, and thus downstream processing (e.g., input to a large language model) can ignore the parenthesized terms for interpretation purposes. For example, the adjusted representation corresponds to "call [him], I mean, send [him] a message." Therefore, the referential resolution has flagged ambiguous terms to indicate that they will be corrected for downstream processing purposes.

[0244] Alternatively, the representation resolution system can adjust the representation by changing ambiguous terms to expected terms, such that the adjusted representation corresponds to "call [John], I mean, message [John]" or "call John, I mean, message John." After adjusting the representation, it is passed to a larger language model for further processing. In another example, the representation is passed to a larger language model without any adjustment, to the extent that entities within the speech representation do not require disambiguation. For example, if the user has already said "call Jane, I mean, message Jane," the representation resolution system can determine that no ambiguous terms were found and can proceed without adjusting the representation before passing it to the larger language model.

[0245] In some cases, a referential resolution system may fail to identify a referent or otherwise resolve ambiguity within a verbal representation. For example, the system may fail to determine the referent for “I mean”. This determination may be based on confidence levels, such that if the system cannot determine the referent for a term with a sufficient threshold confidence level (e.g., 95% confidence level), the system does not resolve the ambiguity. Generally, referential resolution systems utilize device context to resolve ambiguous terms or phrases, such as the context displayed on a screen (e.g., a messaging application associated with identifier 1006). In other cases, the system may utilize undisplayed context, such as recent user activity (e.g., recently used applications, recent contacts, etc.), calendar information, emails, and / or text messages. To the extent that the referential resolution system fails to resolve ambiguity in a verbal representation, the representation is passed to a large language model without adjustment (e.g., without adding appropriate indications). In some implementations, the LLM handles ambiguous terms (those that cannot be resolved by the referencing resolution system) to attempt to rewrite and remove ambiguous words or phrases. For example, a user might say "Take me there" while viewing a specific business or location displayed within the app. Where the referencing resolution system cannot identify the location based on the displayed context, it abandons adjusting the representation and passes the unadjusted representation to the LLM. The LLM can then attempt to resolve the user's request "Take me there" to "Take me to business XYZ".

[0246] Generally, the adjusted representation is used as a factor influencing the representation generated by the language model. Regarding the example above, to the extent that the referential resolution system can resolve "he" to "John," the adjusted representation passed to the LLM includes "call [him], I mean, message [him]." Therefore, the adjusted representation informs the LLM which parts of the speech input (e.g., the word "he") the referential resolution system has resolved. Therefore, the LLM does not rewrite the words marked within the adjusted representation (in this example, using brackets). After the rewriting by the LLM, the representation becomes "message [him]." To the extent that the referential resolution system cannot resolve "he" to a specific contact on the device, the representation passed to the LLM includes "call him, I mean, message him." Therefore, the rewriting by the LLM is not suppressed, allowing the LLM to modify the representation to account for both the user's correction of "I mean" and the ambiguous term "he." Therefore, the rewriting by the LLM becomes "message John."

[0247] There are multiple ways that a referential resolution system and an LLM system can process specific inputs. For example, a user says, “Which country is Khachapri from?” In response, the digital assistant provides the output “Georgia.” The user can then say, “What time is it there?” In one example, the referential resolution system can resolve “there” to “Georgia” based on the context (the country previously output by the digital assistant). In this case, the adjusted representation passed to the LLM includes “[What time is it there]?”, causing the system to output “The time in Tbilisi, Georgia is 5:00 PM.” In different examples, the referential resolution system may not resolve the term “there,” causing the unadjusted representation to be passed to the LLM. Alternatively, some systems may not include a referential resolution system, causing the unadjusted representation to always be passed to the LLM. In these cases, the representation passed to the LLM includes “What time is it there?” The LLM performs a rewrite of the representation and rewrites “there” to “Georgia.” In these cases, downstream systems may incorrectly determine that “Georgia” corresponds to the state of Georgia in the United States. Therefore, the system outputs "The time in Georgia, USA is 1:00 AM". Thus, integrating the referential resolution system with LLM functionality improves these types of systems.

[0248] After the corresponding speech representation is passed to a large language model (adjusted or unadjusted), the large language model processes the representation to determine whether it needs modification (or "rewriting"). For example, if the representation includes grammatical or syntactic errors, user corrections, and / or highly ambiguous words or phrases, modification may be necessary. In this example, the representation received at the large language model includes "call [him], I mean, message [him]". The word "I mean" is identified as the entity that needs modification. In this case, "I mean" corresponds to user correction from one intention to another, such as the intention to "make a call" being corrected to the intention to "make a text message". Therefore, based on the determination that the adjusted representation needs modification, the large language model modifies the adjusted representation to obtain the modified and adjusted representation. (See also: ...) Figure 8 and Figure 9The discussion focuses on training large language models on specific datasets to accurately perform specific types of tasks corresponding to a specific domain or otherwise related in one or more ways. For example, a large language model can be trained on a dataset that includes requests from digital assistants in one or more specific or general environments (e.g., using a mobile phone, tablet, headset, etc.). Thus, the large language model can efficiently transform user utterances with intent correction to output corrected utterances that take the corrections into account. In this case, the large language model takes the input "call [him], I mean, message [him]" and provides the output, considering intent correction, specifically "message [him]". In this example, the large language model has removed the first expected command ("call [him]") and also removed the intent that corrected the first expected command ("I mean") to provide a modified and adjusted representation corresponding to the actual user intent of "message [him]". In some cases, the LLM can incorporate the adjusted representation to change the resulting adjusted and modified representation of the output (compared to the case where the LLM had already received the unadjusted representation).

[0249] refer to Figure 10B Once the obtained representation is obtained, device 1002 initiates task flow processing based on the obtained representation. (See also: Regarding...) Figure 10A The resulting representation discussed may include adjusted representations (including adjustments based on referential resolution), modified representations (including modifications based on large language model processing), or both adjusted and modified representations (including both adjustments based on referential resolution and modifications based on large language model processing). For example, device 1002 may process the resulting representation “Message [him]”, which corresponds to “Message John”. Device 1002 may display a prompt 1012 confirming or otherwise indicating the determined intent “Message John”, and further prompt the user “What do you want to say?”. In this way, device 1002 provides the user with the opportunity to review the resulting intent and provide follow-up as needed. In some examples, the prompt includes text reflecting the original user verbal input (“Call him, I mean, message him.”) and / or the determined intent (“Message John”).

[0250] Figures 11A to 11BProcess 1100 for contextual speech interpretation using a large language model is illustrated according to various examples. For example, process 1100 is performed using one or more electronic devices. In some examples, process 1100 is performed using a client-server system (e.g., system 100), and the boxes of process 1100 are divided in any way between the server (e.g., DA server 106) and the client devices. In other examples, the boxes of process 1100 are divided between the server and multiple client devices (e.g., mobile phones and smartwatches). Therefore, although parts of process 1100 are described herein as being performed by a specific device of the client-server system, it should be understood that process 1100 is not limited thereto. In other examples, process 1100 is performed using only client devices (e.g., user device 104) or only multiple client devices. In process 1100, some boxes are optionally combined, the order of some boxes is optionally changed, and some boxes are optionally omitted. In some examples, additional steps may be performed by combining process 1100.

[0251] refer to Figure 11A At box 1102, in some embodiments, the electronic device receives speech input from a user. At box 1104, the electronic device obtains a representation of the speech input. At box 1106, if it is determined, based on (e.g., using an AI process or a generative AI process), that at least one entity within the representation does not require disambiguation, the electronic device provides the representation to a language model. In some embodiments, if it is determined, based on (e.g., using an AI process or a generative AI process), that the representation needs modification, the language model modifies the representation (e.g., using an AI process or a generative AI process), and provides an output based on the modified representation (e.g., using an AI process or a generative AI process). In some embodiments, if it is determined, based on (e.g., using an AI process or a generative AI process), that the representation does not need modification, the electronic device abandons the modification of the representation by the language model (e.g., using an AI process or a generative AI process), and provides an output based on the representation (e.g., using an AI process or a generative AI process). In some implementations, (e.g., using an AI process or a generative AI process) determining whether at least one corresponding entity within a corresponding representation needs disambiguation includes (e.g., using an AI process or a generative AI process) determining whether the referencing disambiguation system is able to obtain a referencing for at least one corresponding entity with a confidence level exceeding a threshold confidence level.

[0252] At box 1108, if at least one entity within the representation needs disambiguation based on (e.g., using an AI process or a generative AI process), the electronic device adjusts the representation by adding an indication to the at least one entity (e.g., using an AI process or a generative AI process). In some embodiments, if at least one entity within the representation needs disambiguation based on (e.g., using an AI process or a generative AI process): the electronic device determines that a reference for the at least one entity can be obtained based on (e.g., using an AI process or a generative AI process), and adjusts the representation by adding an indication to the at least one entity (e.g., using an AI process or a generative AI process). In some embodiments, if it is determined that a reference for the at least one entity cannot be obtained based on (e.g., using an AI process or a generative AI process): the electronic device abandons adjusting the representation by adding an indication to the at least one entity (e.g., using an AI process or a generative AI process), and provides the representation to the language model. In some embodiments, the electronic device (e.g., using an AI process or a generative AI process) identifies contextual information associated with the electronic device, wherein (e.g., using an AI process or a generative AI process) determining whether at least one entity within the representation needs disambiguation is based on the contextual information associated with the electronic device. In some implementations, contextual information includes at least one of displayed content and undisplayed content. In some implementations, based on (e.g., using an AI process or a generative AI process) determining that at least one entity within the representation needs disambiguation and based on the referential resolution setting: the electronic device abandons providing the adjusted representation to the language model; provides the representation to the language model; the language model modifies the representation (e.g., using an AI process or a generative AI process); adjusts the modified representation based on the obtained referentials for at least one entity (e.g., using an AI process or a generative AI process); and provides output based on the adjusted and modified representations.

[0253] refer to Figure 11BAt box 1110, the electronic device provides the adjusted representation to the language model. At box 1112, based on (e.g., using an AI process or a generative AI process) it is determined that the adjusted representation does not require modification, the electronic device provides output based on the adjusted representation. In some embodiments, at box 1114, based on (e.g., using an AI process or a generative AI process) it is determined that the adjusted representation needs modification, the electronic device modifies the adjusted representation via the language model (e.g., using an AI process or a generative AI process). At box 1116, the electronic device provides output based on the modified and adjusted representation. In some embodiments, modifying the adjusted representation by the language model (e.g., using an AI process or a generative AI process) includes (e.g., using an AI process or a generative AI process) correcting the syntax of the adjusted representation. In some embodiments, modifying the adjusted representation by the language model (e.g., using an AI process or a generative AI process) includes (e.g., using an AI process or a generative AI process) correcting grammatical errors within the adjusted representation. In some implementations, determining whether a corresponding representation needs modification (e.g., using an AI process or a generative AI process) includes detecting whether the corresponding representation includes non-fluency (e.g., using an AI process or a generative AI process). In some implementations, determining whether a corresponding representation needs modification includes detecting whether the corresponding representation includes correction (e.g., using an AI process or a generative AI process). In some implementations, determining whether a corresponding representation needs modification includes detecting whether the corresponding representation includes ambiguity exceeding a threshold. In some implementations, before detecting whether a corresponding representation includes ambiguity exceeding a threshold, the electronic device (e.g., using an AI process or a generative AI process) determines that the referencing resolution system cannot obtain a referencing for at least one entity within the corresponding representation based on the ambiguity.

[0254] The above reference Figures 11A to 11B The described operation may optionally be performed by Figures 1 to 4 , Figures 6A to 6B , Figures 7A to 7C , Figure 8 and Figure 9 The components described herein are used to implement the process. For example, the operation of process 1100 may be implemented by one or more of the following: operating system 718, application module 724, I / O processing module 728, STT processing module 730, natural language processing module 732, vocabulary index 744, task flow processing module 736, service processing module 738, media service 120-1, or processors 220, 410, and 704. Those skilled in the art will clearly understand how to implement the process based on... Figures 1 to 4 , Figures 6A to 6B , Figures 7A to 7C, Figure 8 and Figure 9 The components described herein are used to implement other processes.

[0255] According to some specific embodiments, a computer-readable storage medium (e.g., a non-transitory computer-readable storage medium) is provided that stores one or more programs executable by one or more processors of an electronic device, the one or more programs including instructions for performing any of the methods or processes described herein.

[0256] According to some specific embodiments, an electronic device (e.g., a portable electronic device) is provided, which includes components for performing any of the methods or processes described herein.

[0257] According to some specific embodiments, an electronic device (e.g., a portable electronic device) is provided, the electronic device including a processing unit configured to perform any of the methods or processes described herein.

[0258] According to some specific embodiments, an electronic device (e.g., a portable electronic device) is provided, the electronic device including one or more processors and a memory storing one or more programs for execution by the one or more processors, the one or more programs including instructions for performing any of the methods or processes described herein.

[0259] For purposes of explanation, the foregoing description has been given by reference to specific embodiments. However, the illustrative discussion above is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations are possible based on the teachings above. These embodiments were chosen and described in order to best explain the principles of these techniques and their practical application. Others skilled in the art will thus be able to best utilize these techniques and the various embodiments with various modifications suitable for the particular intended use.

[0260] While this disclosure and examples have been fully described with reference to the accompanying drawings, it should be noted that various changes and modifications will become apparent to those skilled in the art. It should be understood that such changes and modifications are considered to be included within the scope of this disclosure and examples as defined by the claims.

[0261] Some implementations described herein may include the use of artificial intelligence and / or machine learning systems (sometimes referred to herein as AI / ML systems). This use may include the collection, processing, labeling, organizing, analyzing, recommending, and / or generating of data. Entities that collect, share, and / or otherwise utilize user data should provide transparency and / or obtain user consent when collecting such data. This disclosure recognizes that the use of data in AI / ML systems can be used to benefit users. For example, the data can be used to train models that can be deployed to improve the performance, accuracy, and / or functionality of applications and / or services. Therefore, the use of data enables AI / ML systems to adapt and / or optimize operations, thereby providing a more personalized, efficient, and / or enhanced user experience. Such adaptation and / or optimization may include customizing content, recommendations, and / or interactions for individual users, as well as simplifying processes and / or implementing more intuitive interfaces. This disclosure also considers further beneficial uses of data in AI / ML systems.

[0262] This disclosure anticipates that, in some implementations, the data used by the AI / ML system will include publicly available data. To protect user privacy, data may be anonymized, aggregated, and / or otherwise processed to remove or, where possible, limit any individual identification. As discussed herein, entities that collect, share, and / or otherwise utilize such data should obtain user consent before collecting such data and / or provide transparency in the process of collecting it. Furthermore, this disclosure anticipates that entities responsible for the use of data (including, but not limited to, data used in connection with AI / ML systems) should strive to comply with robust privacy policies and / or privacy measures.

[0263] For example, such entities may implement and consistently follow strategies and measures deemed to meet or exceed industry standards and regulatory requirements for developing and / or training AI / ML systems. In doing so, efforts should be made to ensure that all intellectual property and privacy considerations are maintained. Training should include measures to protect training data (such as personal information) by providing adequate safeguards against misuse or exploitation. Such strategies and measures should cover all phases of AI / ML system development, training, and use, including data collection, data preparation, model training, model evaluation, model deployment, and ongoing monitoring and maintenance. Transparency and measurability should always be maintained. Such strategies should be readily accessible to users and should be updated as data collection and / or use change. User data should be collected for the entity's lawful and reasonable use and not shared or sold outside of these lawful uses. Furthermore, such collection and sharing should be conducted in a transparent manner to users and / or with their informed consent. Additionally, such entities should consider taking any necessary steps to protect and safeguard access to such data and ensure that other entities with access to the data comply with their privacy policies and procedures. In addition, such entities should be subject to appropriate third-party assessments for transparency purposes to demonstrate their compliance with widely accepted privacy policies and practices. Furthermore, policies and / or practices should be appropriate for the specific types of data collected and / or accessed, and tailored to specific use cases and applicable laws and standards, including jurisdiction-specific considerations.

[0264] In some implementations, the AI / ML system may utilize a model that can be trained (e.g., in a supervised or unsupervised learning manner) using a variety of training data, including data collected using the user's device. This use of user-collected data may be limited to operations on the user's device. For example, model training may be performed locally on the user's device, so that no part of the data is transmitted to another device. In other implementations, model training may be performed using one or more other devices besides the user's device (e.g., a server), but in a privacy-preserving manner, such as through multi-party computation, encrypted by secretly sharing data, or other means, ensuring that user data is not leaked to these other devices.

[0265] In some implementations, the trained model can be centrally stored on the user device or on multiple devices, as is the case in federated learning. This distributed storage can also be done in a privacy-preserving manner, for example, through encryption, where each piece of data is fragmented so that the data cannot be reassembled or used by a device alone (i.e., only in conjunction with another device) or only by the user device. In this way, user or device behavior patterns are not leaked, while leveraging the increased computing resources of other devices to train and execute the ML model. Therefore, user-collected data can be protected. In some specific implementations, data from multiple devices can be combined in a privacy-preserving manner to train the ML model.

[0266] In some implementations, this disclosure contemplates that data used for AI / ML systems may be kept strictly separate from the platforms used to deploy the AI / ML systems and / or for interacting with and / or processing data with users. In such implementations, data used for offline training of the AI / ML systems may be kept in a secure data repository with restricted access and / or not retained for longer than necessary for the training purposes. In some implementations, AI / ML systems may utilize local memory caching to temporarily store data during user sessions. Local memory caching can be used to improve the performance of the AI / ML systems. However, to protect user privacy, data stored in local memory caching may be erased after the user session ends. Any temporary cached data used for online learning or inference may be quickly erased after processing. All data collection, transmission, and / or storage should be conducted using industry-standard encryption and / or secure communication.

[0267] In some implementations, as noted above, techniques such as federated learning, differential privacy, secure hardware components, homomorphic encryption and / or multi-party computation, and others, can be used to further protect personal information data during the training and / or use of AI / ML systems. AI / ML systems should be monitored for changes in the underlying data distribution, such as concept drift or data skew that may degrade AI / ML system performance over time.

[0268] In some implementations, a combination of offline and online training is used to train the AI / ML system. Offline training can use a carefully selected dataset to establish baseline model performance, while online training allows the AI / ML system to continuously adapt and / or improve. This disclosure recognizes the importance of maintaining strict data management measures throughout this process to ensure user privacy is protected.

[0269] In some implementations, AI / ML systems may be designed with safeguards to maintain adherence to their original intended purpose, even when the AI / ML system is adapted based on new data. Any significant changes to data collection and / or application in the use of the AI / ML system may (and in some cases should) be transparently communicated to affected stakeholders, including (or including) obtaining user consent when the manner of collecting and / or using user data changes.

[0270] Regardless of the foregoing, this disclosure also contemplates implementation schemes for users to selectively restrict and / or block the use and / or access to data. That is, this disclosure contemplates that hardware and / or software components may be provided to prevent or block access to data. For example, with respect to some services, the technology should be configured to allow users to opt-in or opt-out to participate in data collection at any time during or after service registration. In another example, the technology should be configured to allow users to opt out of providing certain data used for training AI / ML systems and / or as input during the inference phase of such systems. In yet another example, the technology should be configured to allow users to choose to limit the length of time data is retained or to completely prohibit AI / ML systems from using their data. In addition to providing "opt-in" and "opt-out" options, this disclosure also contemplates providing notifications related to access to or use of personal information. For example, users may be notified when their data is input into an AI / ML system for training or inference purposes, and / or alerted when the AI / ML system generates output or makes decisions based on their data.

[0271] This disclosure recognizes that AI / ML systems should incorporate explicit limitations and / or oversight to mitigate risks that may still exist even when such systems have been designed, developed, and / or operated in accordance with industry best practices and standards. For example, outputs may be generated that could be considered erroneous, harmful, offensive, and / or biased; such outputs may not necessarily reflect the opinions or positions of the entity that developed or deployed these systems. Furthermore, in some cases, references to third-party products and / or services in these outputs should not be construed as endorsement or affiliation of the third-party products and / or services by the entity providing the AI / ML system. The generated content may be filtered to remove potentially inappropriate or dangerous material before being presented to users, while maintaining the ability for human oversight and / or to cover or correct erroneous or undesirable outputs as a safeguard.

[0272] This disclosure also anticipates that users of the AI / ML system should avoid using the service in any way that infringes upon, misappropriates, or violates the rights of any party. Furthermore, the AI / ML system should not be used for any illegal or unlawful activities, nor should it be used to develop any application or use case that will commit or assist in the committing of a crime or other tortious, illegal, or unlawful act. The AI / ML system should not violate, misappropriate, or infringe upon any party's copyright, trademark, privacy and portrait rights, trade secrets, patents, or other proprietary or statutory rights, and should be appropriately attributed to the content as required. In addition, the AI / ML system should not interfere with any security, digital signature, digital rights management, content protection, verification, or authentication mechanisms. The AI / ML system should not misrepresent machine-generated output as human-generated.

[0273] As described above, one aspect of this technology is the collection and use of data available from various sources to perform contextual speech interpretation using large language models. This disclosure anticipates that, in some instances, such collected data may include personal information data that uniquely identifies or can be used to contact or locate specific individuals. This personal information data may include demographic data, location-based data, telephone numbers, email addresses, Twitter IDs, home addresses, data or records related to a user's health or fitness level (e.g., vital sign measurements, medication information, exercise information), date of birth, or any other identifying or personal information.

[0274] This disclosure recognizes that the use of such personal information data in this technology can benefit users. For example, personal information data can be used to provide modified and adjusted representations. Therefore, the use of such personal information data enables users to have computational control over the modified and adjusted representations. Furthermore, this disclosure also anticipates other uses of personal information data that benefit users. For example, health and fitness data can be used to provide insights into a user's overall health status or can be used as positive feedback for individuals using the technology to pursue health goals.

[0275] This disclosure anticipates that entities responsible for the collection, analysis, disclosure, transmission, storage, or other use of such personal information data will comply with robust privacy policies and / or privacy measures. Specifically, such entities should implement and adhere to privacy policies and measures that are recognized as meeting or exceeding industry or governmental requirements for maintaining the privacy and security of personal information data. Such policies should be easily accessible to users and should be updated as the collection and / or use of data changes. Personal information from users should be collected for legitimate and reasonable entity purposes and should not be shared or sold outside of these legitimate purposes. Furthermore, such collection / sharing should be conducted only after receiving informed consent from users. Additionally, such entities should consider taking any necessary steps to protect and safeguard the right to access such personal information data and ensure that other entities with access to personal information data comply with the privacy policies and procedures of other entities. Furthermore, such entities may subject themselves to third-party assessments to demonstrate their compliance with widely accepted privacy policies and privacy measures. Moreover, policies and measures should be appropriate to the specific types of personal information data collected and / or accessed, and should be compatible with applicable laws and standards, including considerations of specific jurisdictions. For example, in the United States, the collection or acquisition of certain health data may be governed by federal and / or state laws, such as the Health Insurance Portability and Accountability Act (HIPAA); while in other countries, health data may be subject to other regulations and policies and should be handled accordingly. Therefore, different privacy measures should be advocated for different types of personal data in each country.

[0276] Regardless of the foregoing, this disclosure also anticipates implementation schemes for users to selectively block the use or access to personal information data. That is, this disclosure anticipates providing hardware and / or software components to prevent or block access to such personal information data. For example, in the case of contextual speech interpretation using a large language model, this technology can be configured to allow users to opt-in or opt-out at any time during or after registration for the service to participate in the collection of personal information data. In another example, users can choose not to provide personal information (e.g., message data) for task execution and recommendations. In yet another example, users can choose to limit the amount of personalized data obtained (e.g., providing only "available" and "unavailable" indicators for calendar data). In addition to providing "opt-in" and "opt-out" options, this disclosure also anticipates providing notifications related to access to or use of personal information. For example, users may be notified when downloading an application that their personal information data will be accessed, and then reminded again just before the application accesses the personal information data.

[0277] Furthermore, the intent of this disclosure is that personal information data should be managed and processed in a manner that minimizes the risk of unintentional or unauthorized access or use. Once data is no longer needed, this risk can be minimized by restricting data collection and deleting data. Additionally, and where applicable, including in certain health-related applications, data deidentification can be used to protect user privacy. Where appropriate, deidentification can be facilitated by removing specific identifiers (e.g., date of birth, etc.), controlling the amount or characteristics of stored data (e.g., collecting location data at the city level rather than address level), controlling how data is stored (e.g., aggregating data among users), and / or other methods.

[0278] Therefore, while this disclosure broadly covers the use of personal information data to implement one or more of the various disclosed embodiments, it also contemplates that various embodiments can be implemented without access to such personal information data. That is, various embodiments of the present technology will not become inoperable due to the absence of all or part of such personal information data. For example, preferences can be inferred from non-personal information data or an absolute minimum of personal information (such as content requested by a device associated with the user, other non-personal information or publicly available information available for tasks and suggestion architectures) to provide users with contextual speech interpretation using a large language model.

Claims

1. A computer-implemented method, the computer-implemented method comprising: In electronic devices that have memory and one or more processors: Receive verbal input from the user; Obtain a representation of the spoken input; Based on the determination that at least one entity within the representation needs to be disambiguated: The representation is adjusted by adding an indication to the at least one entity; as well as The adjusted representation is provided to the language model; Based on the determined adjustment, the following modifications are required: The representation modified and adjusted by the language model; as well as Output is provided based on the modified and adjusted representation.

2. The method of claim 1, wherein determining that the adjusted representation needs modification occurs after the adjusted representation is provided to the language model.

3. The method according to any one of claims 1 to 2, wherein the method comprises: The representation is provided to the language model based on the determination that at least one entity within the representation does not require disambiguation.

4. The method according to claim 3, wherein the method comprises: Based on the determination that the representation needs to be modified, the language model modifies the representation; as well as The output is provided based on the modified representation.

5. The method according to claim 3, wherein the method comprises: Based on the determination, the representation does not require modification: Abandoning the modification of the representation by the language model; as well as The output is provided based on the representation.

6. The method according to any one of claims 1 to 2, wherein the method comprises: Based on the determination that at least one entity within the representation needs to be disambiguated: Based on the determination that a reference for the at least one entity can be obtained, the representation is adjusted by adding the indication to the at least one entity.

7. The method according to claim 6, wherein the method comprises: Based on the determination that no reference can be obtained for the at least one of the entities: Abandoning the adjustment of the representation by adding the instruction to the at least one entity; as well as The representation is provided to the language model.

8. The method according to any one of claims 1 to 7, the method comprising: Identify contextual information associated with the electronic device, wherein determining whether at least one entity within the representation needs disambiguation is based on the contextual information associated with the electronic device.

9. The method of claim 8, wherein the context information includes at least one of displayed content and undisplayed content.

10. The method according to any one of claims 1 to 9, wherein determining whether at least one corresponding entity within a corresponding representation needs disambiguation includes determining whether the referencing disambiguation system is capable of obtaining a referencing for the at least one corresponding entity at a confidence level exceeding a threshold confidence level.

11. The method according to any one of claims 1, 2, and 6 to 8, wherein the method comprises: Based on the determination that at least one entity within the representation needs disambiguation and based on the dereference resolution settings: Abandon providing the adjusted representation to the language model; The representation is provided to the language model; The representation is modified by the language model; The modified representation is adjusted based on the reference obtained for at least one entity; as well as Output is provided based on the adjusted and modified representation.

12. The method according to any one of claims 1, 2, and 6 to 8, wherein the method comprises: The output is provided based on the adjusted representation, provided that the adjusted representation is determined to be unchanged.

13. The method according to any one of claims 1 to 12, wherein determining whether a corresponding representation needs modification includes detecting whether the corresponding representation includes non-fluency.

14. The method according to any one of claims 1 to 13, wherein determining whether a corresponding representation needs modification includes detecting whether the corresponding representation includes correction.

15. The method according to any one of claims 1 to 14, wherein determining whether a corresponding representation needs to be modified includes detecting whether the corresponding representation includes ambiguity exceeding a threshold ambiguity.

16. The method of claim 15, wherein before detecting whether the corresponding representation includes ambiguity exceeding a threshold ambiguity, it is determined that the referencing resolution system cannot obtain a referencing for at least one entity within the corresponding representation based on the ambiguity.

17. The method according to any one of claims 1 to 16, wherein modifying the adjusted representation by the language model includes correcting the syntax of the adjusted representation.

18. The method according to any one of claims 1 to 17, wherein modifying the adjusted representation by the language model includes correcting grammatical errors within the adjusted representation.

19. An electronic device, the electronic device comprising: One or more processors; Memory; and One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising instructions for: Receive verbal input from the user; Obtain a representation of the spoken input; Based on the determination that at least one entity within the representation needs to be disambiguated: The representation is adjusted by adding an indication to the at least one entity; as well as The adjusted representation is provided to the language model; Based on the determined adjustment, the following modifications are required: The representation modified and adjusted by the language model; as well as Output is provided based on the modified and adjusted representation.

20. A non-transitory computer-readable storage medium storing one or more programs configured to be executed by one or more processors of an electronic device, the one or more programs comprising instructions for: Receive verbal input from the user; Obtain a representation of the spoken input; Based on the determination that at least one entity within the representation needs to be disambiguated: The representation is adjusted by adding an indication to the at least one entity; as well as The adjusted representation is provided to the language model; Based on the determined adjustment, the following modifications are required: The representation modified and adjusted by the language model; as well as Output is provided based on the modified and adjusted representation.

21. An electronic device, the electronic device comprising: Components used to detect event alarms; Components for enabling the provision of a message at a second electronic device, wherein the message is associated with the event alarm; A component for receiving motion data corresponding to the movement of the second electronic device from the second electronic device; Components used to determine posture based on the motion data; Components for providing a first output at the second electronic device based on the posture; and A component used to perform the first task associated with the event alarm.

22. An electronic device, the electronic device comprising: One or more processors; Memory; and One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing the method according to any one of claims 1 to 20.

23. A non-transitory computer-readable storage medium storing one or more programs, said one or more programs comprising instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform the method according to any one of claims 1 to 20.

24. An electronic device, the electronic device comprising: Components for performing the method according to any one of claims 1 to 20.

25. A computer program product comprising one or more programs configured to be executed by one or more processors of an electronic device, the one or more programs comprising instructions for performing the method according to any one of claims 1 to 20.

Images