Electronic device and voice recognition method thereof

Abstract

According to an embodiment, an electronic device includes: a microphone that receives a user's voice; a memory that stores a plurality of voice recognition models; and a processor that selects a voice recognition model from the plurality of voice recognition models stored in the memory based on an operating state of the electronic device, and recognizes the user's voice received by the microphone based on the selected voice recognition model.

Description

Technical Field

[0001] The present disclosure relates to an electronic device capable of recognizing a user's voice and a voice recognition method thereof. Background Art

[0002] Recently, in order to increase user convenience, voice recognition services for recognizing a user's voice when the user utters his / her desired command and detecting a command included in the user's utterance to thereby provide a service corresponding to the user's intention have been developed and provided.

[0003] On the other hand, in order to improve the speech recognition rate and speech recognition accuracy, it is important to measure the noise of the environment in which the speech recognition service is provided. As the noise of the environment in which the speech recognition service is provided, the noise generated by the device providing the speech recognition service may have a great influence on the speech recognition performance. Summary of the invention

[0004] Description of Related Technology

[0005] Aspects of the present disclosure address at least the above-mentioned problems and / or disadvantages and provide at least the advantages described below.Accordingly, one aspect of the present disclosure provides an electronic device capable of recognizing a user's voice and a voice recognition method thereof.

[0006] Technical Solution

[0007] According to one aspect of the present disclosure, an electronic device is provided. The electronic device includes: a microphone configured to receive a user's voice; a memory storing a plurality of voice recognition models; and at least one processor configured to select a voice recognition model from the plurality of voice recognition models stored in the memory based on an operation state of the electronic device, and recognize the user's voice received by the microphone based on the selected voice recognition model.

[0008] The operation state may include at least one of an operation mode or an operation intensity of the electronic device.

[0009] At least one processor may be further configured to select a speech recognition model corresponding to a state condition satisfying an operating state among a plurality of state conditions as a speech recognition model corresponding to the operating state, wherein the plurality of state conditions are respectively matched with the plurality of speech recognition models.

[0010] The at least one processor may be further configured to, when the operating state changes, change the selected speech recognition model based on the changed operating state, and recognize the user's speech based on the changed speech recognition model.

[0011] The at least one processor may be further configured to determine a sensitivity of the selected speech recognition model based on the operating state, and recognize the user's speech based on the selected sensitivity.

[0012] The at least one processor may be further configured to, when the operating state changes, change the sensitivity based on the changed operating state.

[0013] The at least one processor may be further configured to determine the user's intention based on a result of speech recognition of the user's voice, and to change the sensitivity based on the user's intention.

[0014] The at least one processor may be further configured to determine a first pattern for noise of the surrounding environment based on the operating state, and generate a speech recognition model based on the determined first pattern.

[0015] The at least one processor may be further configured to determine a second pattern for noise of the electronic device based on the operating state, and determine the first pattern for noise of the surrounding environment based on the second pattern.

[0016] The multiple speech recognition models may include multiple speech recognition models for recognizing wake-up words.

[0017] The electronic device may include at least one of a cleaner, an air conditioner, a refrigerator, a washing machine, or a clothing care appliance.

[0018] According to one aspect of the present disclosure, an electronic device is provided. The electronic device includes: a microphone configured to receive a user's voice; a transceiver configured to communicate with a server; and at least one processor configured to select a voice recognition model from a plurality of voice recognition models received from the server based on an operating state of the electronic device, and recognize the user's voice received by the microphone based on the selected voice recognition model.

[0019] According to another aspect of the present disclosure, a speech recognition method of an electronic device is provided, the method comprising: receiving a user's speech; selecting a speech recognition model from a plurality of pre-stored speech recognition models based on an operation state of the electronic device; and recognizing the user's speech based on the selected speech recognition model.

[0020] Selecting a speech recognition model may include: selecting a speech recognition model corresponding to a state condition satisfying an operating state among a plurality of state conditions as a speech recognition model corresponding to the operating state, wherein the plurality of state conditions are respectively matched with the plurality of speech recognition models.

[0021] The speech recognition method may further include, when the operating state changes, changing the selected speech recognition model based on the changed operating state, wherein recognizing the user's speech may include recognizing the user's speech based on the changed speech recognition model.

[0022] The speech recognition method may further include determining a sensitivity of the selected speech recognition model based on an operation state, wherein recognizing the user's speech based on the selected speech recognition model may include recognizing the user's speech based on the determined sensitivity.

[0023] The speech recognition method may further include: when the operating state changes, changing the sensitivity of the selected speech recognition model based on the changed operating state, wherein recognizing the user's speech based on the selected speech recognition model may include recognizing the user's speech based on the changed sensitivity.

[0024] The voice recognition method may further include: determining the user's intention based on a result of the voice recognition of the user's voice; and changing the sensitivity based on the user's intention.

[0025] The speech recognition method may further include: determining a first pattern for noise of the surrounding environment based on the operation state; and generating a speech recognition model based on the determined first pattern.

[0026] Other aspects, advantages, and salient features of the disclosure will become apparent to those skilled in the art from the following detailed description, which, taken in conjunction with the annexed drawings, discloses various embodiments of the disclosure.

[0027] Advantageous Effects of the Invention

[0028] In an electronic device and a speech recognition method thereof according to one aspect, since speech recognition is performed by a speech recognition model reflecting an ambient noise environment including noise generated in the electronic device, a speech recognition rate and speech recognition accuracy may be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A speech recognition system according to an embodiment of the present disclosure is shown;

[0030] Figure 2 is a control block diagram of an electronic device according to a first embodiment of the present disclosure;

[0031] Figure 3 An example of a state condition table for selecting a speech recognition model in an electronic device according to a first embodiment of the present disclosure is shown;

[0032] Figure 4 is a control block diagram of an electronic device according to a second embodiment of the present disclosure;

[0033] Figure 5 shows the appearance of a robot cleaner as an implementation example of the electronic device according to the second embodiment of the present disclosure;

[0034] Figure 6shows a bottom portion of a robot cleaner as an implementation example of the electronic device according to the second embodiment of the present disclosure;

[0035] Figure 7 An example of a state condition table for selecting a speech recognition model in an electronic device according to a second embodiment of the present disclosure is shown;

[0036] Figure 8 is a control block diagram of an electronic device according to a third embodiment of the present disclosure;

[0037] Fig. 9 shows the appearance of an air conditioner as an implementation example of the electronic device according to the third embodiment of the present disclosure;

[0038] Fig.10 is an exploded perspective view of an air conditioner as an implementation example of the electronic device according to the third embodiment of the present disclosure; and

[0039] Fig.11 , Fig.12 and Fig.13 is a flowchart illustrating a voice recognition method of an electronic device according to various embodiments of the present disclosure. DETAILED DESCRIPTION

[0040] The following description with reference to the accompanying drawings is provided to help fully understand the various embodiments of the present disclosure as defined by the claims and their equivalents. It includes various specific details to help understanding, but these should be considered as merely exemplary. Therefore, those of ordinary skill in the art will recognize that various changes and modifications may be made to the various embodiments described herein without departing from the scope of the present disclosure. In addition, for the sake of clarity and brevity, the description of well-known functions and structures may be omitted. The terms and words used in the following description and claims are not limited to bibliographic meanings, but are only used by the inventors to enable a clear and consistent understanding of the present disclosure. Therefore, it should be clear to those skilled in the art that the following description of the various embodiments of the present disclosure is provided only for the purpose of illustration, not for limiting the present disclosure as defined by the attached claims and their equivalents. Hereinafter, throughout this specification, the same reference numerals will refer to the same components. This specification does not describe all components of the embodiments, and will not describe general information in the technical field to which the present disclosure belongs or overlapping information between the embodiments. As used herein, the terms "part", "section", "module", "component" or "block" may be implemented as software or hardware, and depending on the implementation, multiple "parts", "sections", "modules", "components" or "blocks" may be implemented as a single component, or a single "part", "section", "module", "component" or "block" may include multiple components.

[0041] Furthermore, it will be understood that when a certain part “includes” a certain component, the part does not exclude another component but may further include another component unless the context clearly states otherwise.

[0042] Furthermore, it will be understood that the terms “includes,” “comprises,” “including” and / or “comprising,” when used in the present specification, specify the presence of stated components but do not preclude the presence or addition of one or more other components.

[0043] Throughout the specification, it will also be understood that when an element is referred to as being “on” or “over” another element, it can be directly on the other element or intervening elements may also be present.

[0044] In addition, it will be understood that although the terms "first", "second", etc. may be used herein to describe various components, these components should not be limited by these terms. These terms are only used to distinguish one component from another component.

[0045] It is to be understood that the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise.

[0046] Reference signs used in the operations are provided for convenience of description and do not describe the order of the operations, and the operations may be performed in an order different from that set forth unless a particular order is clearly specified in the context.

[0047] Hereinafter, the operation principle and embodiments of the present disclosure will be described with reference to the accompanying drawings.

[0048] An electronic device according to one aspect may be a device capable of receiving sound through a microphone and transmitting and receiving data through communication with an external device. For example, the electronic device may be a household appliance such as a robot cleaner, an air conditioner, a refrigerator, a washing machine, a clothing care device, an air cleaner, a humidifier, an oven, a microwave oven, an audio system, a television (TV), a speaker, a computer, etc., or may be a mobile device such as a smart phone, a tablet personal computer (PC), a PC, a digital assistant (PDA), etc.

[0049] In addition, the electronic device according to one aspect may perform voice recognition or provide a service provided by voice recognition, and according to some embodiments may perform machine learning or provide the result of machine learning. Therefore, a cleaner, air conditioner, refrigerator, television, speaker, etc. as an electronic device may also be referred to as an artificial intelligence (AI) cleaner, AI air conditioner, AI refrigerator, AI television, AI speaker, etc., respectively.

[0050] The electronic device according to one aspect can perform operations to be described below, and there is no limitation regarding its type, implementation method, name, etc.

[0051] Figure 1 A speech recognition system according to an embodiment of the present disclosure is shown.

[0052] Reference Figure 1 , a user of the voice recognition system 1 according to an embodiment of the present disclosure may speak a control command to the electronic device 100 to input the control command to the electronic device 100 .

[0053] The electronic device 100 may recognize the user's voice and output a response corresponding to the user's voice.

[0054] After the electronic device 100 recognizes the user's voice, the electronic device 100 may output a response corresponding to the user's intention as a response corresponding to the user's voice. At this time, the electronic device 100 may generate a control command or a conversation response corresponding to the user's intention and output the control command or the conversation response.

[0055] On the other hand, the electronic device 100 may be implemented as a home appliance providing various functions, such as a robot cleaner 100a, an air conditioner 100b, a refrigerator 100c, a washing machine 100d, an oven 100e, a clothing care apparatus 100f, and the like.

[0056] The operating state of the electronic device 100 may be a factor affecting the voice recognition performance for the user's voice. The reason may be because the noise of the electronic device 100 generated according to the operating state of the electronic device 100 may act as a factor interfering with accurate voice recognition.

[0057] In addition, since the electronic device 100 provides various functions, the operating state of the electronic device 100 may change every minute. When the operating state of the electronic device 100 changes, the noise generated in the electronic device 100 may also change. Therefore, in order to improve the accuracy of speech recognition, it may be necessary to use a speech recognition model that appropriately responds to changes in environmental noise including noise generated in the electronic device 100.

[0058] Figure 2 is a control block diagram of an electronic device according to a first embodiment of the present disclosure, and Figure 3 An example of a state condition table for selecting a speech recognition model in an electronic device according to the first embodiment of the present disclosure is shown.

[0059] Reference Figure 2 and Figure 3, the electronic device 100 according to the first embodiment of the present disclosure may include a microphone 110 for receiving a user's voice, a communicator or a transceiver 120 , a processor 130 , a memory 140 , and a driver 150 .

[0060] The microphone 110 may receive sound and convert the sound into an electrical signal. A single microphone 110 may be provided in the electronic device 100, or a plurality of microphones 110 may be provided to improve voice recognition performance.

[0061] The microphone 110 may be mounted on an outer surface of the electronic device 100, or may be physically separated from the electronic device 100 to be positioned close to the user. For example, the microphone 110 may be implemented as a movable stand microphone, a wearable microphone (such as a headset), etc. However, as long as the microphone 110 receives the user's voice, the installation position or implementation type of the microphone 110 is not limited.

[0062] The voice spoken by the user may be converted into an electric signal through the microphone 110 and input to the processor 130. Hereinafter, the voice converted into the electric signal is also referred to as a voice signal.

[0063] The transceiver 120 may include a communication circuit for data exchange with an external device or for data exchange between components included in the electronic device 100 .

[0064] The transceiver 120 may transmit / receive various information to / from a server (not shown). The transceiver 120 may receive a speech recognition model from a server (not shown) and store the received speech recognition model in the memory 140.

[0065] To this end, the transceiver 120 may include at least one communication module for transmitting / receiving data according to a predefined communication standard. For example, the transceiver 120 may include at least one of a short-range communication module, a wired communication module, or a wireless communication module.

[0066] The short-range communication module may include various short-range communication modules, such as a Bluetooth module, an infrared communication module, a radio frequency identification (RFID) communication module, a wireless local area network (WLAN) communication module, a near field communication (NFC) module, a Zigbee communication module, etc., to send / receive signals at a short distance through a wireless communication network.

[0067] The wired communication module may include various cable communication modules and various wired communication modules, cable communication modules such as Universal Serial Bus (USB), High-Definition Multimedia Interface (HDMI), Digital Visual Interface (DVI), Recommended Standard-232 (RS-232), power line communication or plain old telephone service (POTS), and wired communication modules such as Controller Area Network (CAN) communication module, local area network (LAN) module, wide area network (WAN) module, or value added network (VAN) module.

[0068] The wireless communication module may include a wireless communication module supporting various wireless communication methods, such as Global System for Mobile Communications (GSM), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Universal Mobile Telecommunications System (UMTS), Time Division Multiple Access (TDMA), Long Term Evolution (LTE), etc., and may include a Wireless Fidelity (WiFi) module and a wireless broadband module.

[0069] The wireless communication module may include such a wireless communication interface, which includes an antenna and a transmitter for transmitting a signal. In addition, the wireless communication module may also include a call conversion module, which is used to convert a digital control signal output from the processor through the wireless communication interface into an analog wireless signal according to the control of the controller.

[0070] The wireless communication module may include a wireless communication interface including an antenna and a receiver for receiving signals. In addition, the wireless communication module may also include a signal conversion module for demodulating an analog wireless signal received through the wireless communication interface into a digital control signal.

[0071] The memory 140 may store various data related to the operation of the electronic device 100 and store data related to voice recognition.

[0072] The memory 140 may store a plurality of speech recognition models 141 ( 141 - 1 , 141 - 2 , . . . 141 - n ) The memory 140 may store a plurality of speech recognition models 141 corresponding to various operation states of the electronic device 100 , respectively.

[0073] The speech recognition model 141 may be received from an external server (not shown) through the transceiver 120, but is not limited thereto. However, the speech recognition model 141 may be pre-stored in the memory 140 at the time of design.

[0074] To this end, the memory 140 may be implemented as at least one of a nonvolatile memory device (e.g., a cache, a read-only memory (ROM), a programmable ROM (PROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), and a flash memory), a volatile memory device (such as a random access memory (RAM)), or a storage medium (such as a hard disk drive (HDD) and a compact disk read-only memory (CD-ROM)). But not limited thereto. The memory 140 may be implemented as a chip separate from the processor 130 to be described later, or the memory 140 and the processor 130 may be integrated into a single chip.

[0075] The driver 150 may include at least one component for generating a driving force or transmitting a driving force for enabling the electronic device 100 to provide at least one function.

[0076] The driver 150 may include various components according to implementation examples of the electronic device 100. For example, the driver 150 may include a motor and, according to some cases, a fan. The driver 150 will be described in detail later.

[0077] The electronic device 100 may include a processor 130 electrically connected to a microphone 110 , a transceiver 120 , a memory 140 , and a driver 150 .

[0078] The processor 130 may recognize the user's voice received through the microphone 110 .

[0079] More specifically, the processor 130 may output the speech in the form of text based on the voice signal delivered from the microphone 110 to recognize the user's voice as a sentence. To this end, the processor 130 may include a voice recognition engine.

[0080] For example, the processor 130 may detect an actual speech segment included in the input speech through endpoint detection (EPD), and extract a feature vector of the input speech from the detected actual speech segment. At this time, the processor 130 may apply a feature vector extraction technique such as cepstrum, linear prediction coefficient (LPC), Mel-frequency cepstrum coefficient (MFCC), or filter bank energy to the detected actual speech segment to extract a feature vector of the input speech.

[0081] The processor 130 may compare the extracted feature vector with the trained reference pattern to obtain a recognition result. To this end, the processor 130 may use the speech recognition model 141 stored in the memory 140.

[0082] The speech recognition model 141 may include at least one of an acoustic model or a language model, wherein the acoustic model is used to model and compare signal features of speech, and the language model is used to model the language order relationship of words, syllables, etc. corresponding to the recognition vocabulary. Alternatively, the speech recognition model 141 may be a model that integrates the acoustic model and the language model.

[0083] Acoustic models can be divided into direct comparison methods and statistical model methods. The direct comparison method sets the recognition target as a feature vector model and compares the feature vector model with the feature vector of the speech data, while the statistical model method performs statistical processing and use on the feature vector of the recognition target.

[0084] The direct comparison method is a method of setting a unit such as a word, a phoneme, etc. as a recognition target as a feature vector model and determining the similarity between the input speech and the feature vector model. A representative example of the direct comparison method is the vector quantization method. The vector quantization method is a method of mapping the feature vector of the input speech data onto a codebook (which is a reference model encoded as a representative value) and comparing the encoded values ​​with each other.

[0085] The statistical model method is a method of configuring the unit for identifying the target as a state sequence and using the relationship between the state sequences. The state sequence can be configured with multiple nodes. The method of using the relationship between the state sequences can include dynamic time warping (DTW), hidden Markov model (HMM), a method using artificial neural network (ANN), etc.

[0086] Dynamic Time Warping (DTW) is a method of compensating for differences on the time axis when comparing with a reference model by taking into account the dynamic characteristics of speech in which the signal length changes over time even when the same person speaks the same word, and the Hidden Markov Model (HMM) is a recognition technology that, after assuming that speech is a Markov process with a state change probability and a node (output symbol) observation probability in each state, estimates the state change probability and the observation probability of the node through learning data, and calculates the probability that the input speech will be generated in the estimation model.

[0087] On the other hand, a language model for modeling a language order relationship of words, syllables, etc. reduces acoustic ambiguity and recognition errors by applying the order relationship between units constituting the language to the units obtained from speech recognition. As language models, there are statistical language models and models based on finite state automata (FSA), and the statistical language model uses chain probabilities of words, such as Unigram, Bigram, Trigram, etc.

[0088] The speech recognition model recognizes speech using any of the methods described above. For example, the processor 130 may use a speech recognition model including an acoustic model applying a hidden Markov model (HMM), or a speech recognition model using an N-best search method that integrates an acoustic model and a language model.

[0089] In addition, the processor 130 can calculate the confidence level to ensure the reliability of the recognition result. The confidence level is a measure representing how reliable the speech recognition result is. For example, the confidence level of a phoneme or word as a recognition result can be defined as a relative value of the probability that the corresponding phoneme or word is spoken from different phonemes or words. Therefore, the confidence level can be a value in the range from 0 to 1 or a value in the range from 0 to 100. When the confidence level of the recognition result exceeds a predefined threshold, the recognition result can be output to perform an operation corresponding to the recognition result, and, when the confidence level of the recognition result is less than or equal to a predefined threshold, the recognition result can be rejected.

[0090] On the other hand, the processor 130 can recognize a predefined word. In this case, the processor 130 can determine whether the recognition result matches the predefined word. When the ratio of the match between the recognition result and the predefined word is greater than or equal to a predefined reference value, the processor 130 can determine that the corresponding word has been spoken by the user.

[0091] For example, the processor 130 may recognize a predefined wake-up word. The wake-up word may be a control command for enabling the voice recognition mode.

[0092] The processor 130 may recognize the wake-up word by using a pre-stored speech recognition model. In this case, the pre-stored speech recognition model may have a recognition range for recognizing predefined words. For example, the speech recognition model may be a speech recognition model for recognizing the wake-up word.

[0093] When the recognition range is limited to a predefined wake-up word, the capacity of the memory 140 or the processor 130 required for voice recognition can be reduced, but is not limited thereto. However, the processor 130 can recognize more languages.

[0094] Furthermore, the processor 130 may apply natural language understanding (NLU) to the utterance in text form as the recognition result to understand the user's intention included in the utterance.

[0095] The processor 130 may perform morpheme analysis on the utterance in text form and analyze the voice behavior possessed by the utterance. Voice behavior analysis may be a task of analyzing the utterance intention of the user to understand the utterance intention regarding whether the user asks a question, whether the user makes a request, whether the user gives an answer, whether the user simply expresses emotions, etc.

[0096] The processor 130 may output a control command or response corresponding to the user's intention understood through morpheme analysis and voice behavior analysis. The processor 130 may generate a control command for executing a service corresponding to the user's intention, or output a response corresponding to the user's intention.

[0097] On the other hand, a speech recognition model for performing an entity or a portion of the speech recognition process described above may be stored in the memory 140, and a plurality of different speech recognition models may be stored in the memory 140. The plurality of speech recognition models stored in the memory 140 may have different parameters. That is, the plurality of speech recognition models may be models modeled to perform the speech recognition process described above by using different parameters, or models using different reference data.

[0098] A plurality of different speech recognition models can be modeled to be suitable for different environments and stored in the memory 140. Therefore, different speech recognition models can be applied according to the environment in which speech recognition is used to achieve a high speech recognition rate and a high speech recognition accuracy. Therefore, the processor 130 can select a speech recognition model suitable for the environment in which speech recognition is performed. Hereinafter, the operation of selecting a speech recognition model in the processor 130 will be described in detail.

[0099] The processor 130 may select a speech recognition model from a plurality of speech recognition models stored in the memory 140, and recognize the user's voice by using the selected speech recognition model. At this time, the processor 130 may reflect the state of the electronic device 100 that may affect the speech recognition to select a speech recognition model from a plurality of pre-stored speech recognition models.

[0100] More specifically, the processor 130 may select a speech recognition model from a plurality of pre-stored speech recognition models based on the operating state of the electronic device 100. Herein, the operating state of the electronic device 100 may be a state factor that may affect the noise generated in the electronic device 100, that is, may be a state factor from which the noise of the electronic device 100 is estimated.

[0101] The operation state of the electronic device 100 may include at least one of an operation mode or an operation intensity of the electronic device 100 .

[0102] The operation mode may be information representing a function provided by the electronic device 100. For example, the operation mode may be classified into a standby mode or an execution mode, wherein the standby mode represents a state in which the electronic device 100 waits to receive a command from a user, and the execution mode represents a state in which the electronic device 100 operates in response to an operation command received from a user. In addition, the operation mode may be classified into at least one mode representing various functions provided by the electronic device 100. The volume, intensity, pattern, etc. of the noise generated in the electronic device 100 may depend on the operation mode.

[0103] On the other hand, the operation mode may be automatically set according to a control command of the processor 130, or the operation mode may be received from the user. When the operation mode is received from the user, the electronic apparatus 100 may further include an input device (not shown).

[0104] The operation intensity may be information representing the intensity of the driving force generated in the electronic device 100. For example, the operation intensity may be represented as data representing the intensity of the motor, such as the torque, revolutions per minute (rpm), current, etc. of the motor. In addition, the operation intensity may be represented as data representing the rpm of the fan. The volume, intensity, pattern, etc. of the noise generated in the electronic device 100 may depend on the operation intensity.

[0105] In addition to the operation mode and the operation intensity, the operation state may further include at least one of an operation component or an operation position.

[0106] The operating component may represent a component in an operating state among at least one component included in the electronic device 100. For example, the operating component may represent a component in an operating state among at least one component included in the driver 150. When operating, different components may generate noises with different volumes, intensities, patterns, etc.

[0107] The operation position may be the position of the electronic device 100, and may be a position that affects the noise generated in the electronic device 100. The operation position may be classified according to predefined areas. For example, the operation position may be classified as under a bed, near a wall, on a carpet, etc. The waveform of the noise generated during operation may depend on the operation position, and therefore, the volume, intensity, pattern, etc. of the noise may change according to the operation position.

[0108] Based on the operating state described above, the processor 130 can select a speech recognition model corresponding to the operating state from among a plurality of speech recognition models. That is, the processor 130 can select a speech recognition model that can ensure a high speech recognition rate and high recognition performance according to the operating state of the electronic device 100.

[0109] More specifically, the processor 130 can identify a state condition that satisfies the operating state from among the predefined state conditions for a plurality of speech recognition models, and select a speech recognition model corresponding to the state condition that satisfies the operating state as the speech recognition model corresponding to the operating state. The predefined state conditions for a plurality of speech recognition models can be conditions for the operating state, and can be defined at design time or according to user input.

[0110] To this end, a state condition table in which state conditions for operating states are matched with speech recognition models may be stored in the memory 140. The processor 130 may compare the operating state of the electronic device 100 with the state condition table to determine a speech recognition model corresponding to the operating state.

[0111] For example, Figure 3 As shown in , the state condition may be pre-defined as a condition for an operation mode among the operation states, and a state condition table in which a speech recognition model is matched with an operation mode may be stored in the memory 140 .

[0112] In this case, the first speech recognition model, the second speech recognition model, and the third speech recognition model stored in the memory 140 may be modeled to have different speech recognition parameters.

[0113] When the operation mode of the electronic device 100 is the first mode, the processor 130 may select the first speech recognition model from the plurality of speech recognition models. Figure 3 For example, when the operation mode of the electronic device 100 is the second mode, the processor 130 may select the second speech recognition model, and, when the operation mode of the electronic device 100 is the third mode, the processor 130 may select the third speech recognition model.

[0114] In this way, the processor 130 can improve the speech recognition rate and speech recognition accuracy by selecting a speech recognition model suitable for a noisy environment according to the operating state of the electronic device 100 instead of unconditionally using a speech recognition model modeled in a normal environment.

[0115] In particular, when the speech recognition model is a wake-up word recognition model, the wake-up word can be accurately recognized despite the noise of the electronic device 100. Therefore, although the user no longer speaks the wake-up word, the wake-up word can be recognized quickly and accurately.

[0116] On the other hand, when the operating state of the electronic device 100 changes, the noise generated in the electronic device 100 may also change.

[0117] When the operating state of the electronic device 100 changes, the processor 130 may change the selected speech recognition model based on the changed operating state, and recognize the user's voice based on the changed speech recognition model. To this end, the processor 130 may acquire the operating state in real time or at predefined time intervals.

[0118] For example, according to Figure 3 For example, when the first mode is changed to the second mode, the processor 130 may change the first speech recognition mode to the second speech recognition mode. Thereafter, when the second mode is changed to the third mode, the processor 130 may change the second speech recognition model to the third speech recognition model, and perform speech recognition by using the third speech recognition model.

[0119] By changing the voice recognition model according to the operating state of the electronic device 100, the processor 130 can improve the accuracy and efficiency of voice recognition.

[0120] The processor 130 may determine the parameters of the speech recognition model based on the operating state, and perform speech recognition based on the determined parameters. The parameters of the speech recognition model may include sensitivity, and may include various speech recognition related parameters other than sensitivity. In the following, sensitivity will be described as an example of the parameters of the speech recognition model.

[0121] The processor 130 may identify a state condition satisfying the operating state from predefined conditions for the operating state for sensitivity, and determine the sensitivity corresponding to the state condition satisfying the operating state as the sensitivity of the speech recognition model.

[0122] To this end, a sensitivity table in which the sensitivity of the speech recognition model is matched with the operating state may be stored in the memory 140. The processor 130 may compare the operating state of the electronic device 100 with the sensitivity table to determine the sensitivity of the speech recognition model corresponding to the operating state.

[0123] The sensitivity may be classified into a plurality of different predefined levels according to the speech recognition model. The processor 130 may determine the sensitivity level corresponding to the operation state.

[0124] For example, when the electronic device 100 is in the first mode, the processor 130 may determine the sensitivity of the selected speech recognition model to be the first level. When the electronic device 100 is in the second mode that generates a louder noise than the first mode, the processor 130 may determine the sensitivity of the speech recognition model to be the second level. In this article, the second level may have a higher sensitivity than the first level.

[0125] In this way, since the processor 130 determines the sensitivity of the voice recognition model according to the operation state, the processor 130 can reduce the influence generated by the noise of the electronic device 100. Therefore, the accuracy and efficiency of voice recognition can be improved.

[0126] On the other hand, when the operating state of the electronic device 100 changes, the noise generated in the electronic device 100 may also change. Therefore, there may be a situation where the sensitivity of the speech recognition model needs to be changed.

[0127] When the operating state changes, the processor 130 may change the sensitivity based on the changed operating state, and recognize the user's voice based on the changed sensitivity. To this end, the processor 130 may acquire the operating state in real time or at predefined time intervals.

[0128] The operation of changing the sensitivity of the speech recognition model can be performed simultaneously with the operation of changing the speech recognition model. That is, when the operating state changes, the processor 130 can change the speech recognition model and the sensitivity of the speech recognition model based on the changed operating state, but is not limited to this. However, the processor 130 can change the sensitivity of the speech recognition model without changing the speech recognition model.

[0129] On the other hand, the sensitivity of the speech recognition model can be changed based on the user's intention rather than the operating state.

[0130] The processor 130 may determine the user's intention based on a result of voice recognition of the user's voice, and change the sensitivity based on the user's intention.

[0131] More specifically, the processor 130 may understand the context based on the result of the speech recognition, understand the user's intention regarding whether the user will continue speaking or finish speaking according to the context, and change the sensitivity based on the user's intention.

[0132] For example, when the processor 130 determines that the user will continue to speak based on the result of the voice recognition of the user's voice, the processor 130 can change the sensitivity to a higher sensitivity. When the processor 130 determines that the user will end speaking based on the result of the voice recognition of the user's voice, the processor 130 can change the sensitivity to a lower sensitivity.

[0133] In this way, since the sensitivity of the speech recognition model changes according to the user's intention, accurate speech recognition services can be provided in various environments where speech recognition is used.

[0134] In particular, when recognizing a wake-up word, since the sensitivity changes according to the user's intention after the user says the wake-up word, more accurate voice recognition can be achieved. Therefore, the convenience of the user can be increased.

[0135] On the other hand, the processor 130 may generate a speech recognition model for the noise of the surrounding environment based on the operation state, and store the speech recognition model in the memory 140 or a server (not shown).

[0136] The processor 130 may determine a first pattern for ambient noise based on the operating state and generate a speech recognition model based on the first pattern. The ambient noise may be noise excluding the noise of the electronic device 200 from among the noise input to the microphone 110.

[0137] More specifically, the processor 130 may determine a second pattern for noise of the electronic device 100 (ie, noise generated in the electronic device 100 ) based on the operation state, and determine a first pattern for noise of the surrounding environment based on the second pattern.

[0138] To this end, the noise pattern according to the operating state may be stored in the memory 140 or the server (not shown). The processor 130 may determine the noise pattern corresponding to the operating state among the noise patterns stored in the memory 140 or the server (not shown) as the second pattern of noise for the electronic device 100.

[0139] The processor 130 may determine noise excluding the noise of the electronic device 100 of the second pattern among the noise input to the microphone 110 as ambient noise, and extract the first pattern for the ambient noise based on the signal characteristics of the determined ambient noise.

[0140] The processor 130 may generate a speech recognition model based on a first pattern for noise of the surrounding environment.

[0141] Furthermore, the processor 130 may change at least one of a pre-selected speech recognition model or a sensitivity of the speech recognition model based on the first pattern with respect to the noise of the surrounding environment.

[0142] Therefore, despite the presence of noise in the surrounding environment and noise generated in the electronic device, accurate and effective voice recognition can be achieved. Therefore, user convenience can be increased.

[0143] On the other hand, data for an algorithm for controlling the operation of components in the electronic device 100 or a program for implementing the algorithm may be stored in the memory 140. The processor 130 may perform the above-described operations by using the data stored in the memory 140. The memory 140 and the processor 130 may be implemented as separate chips or integrated into a single chip.

[0144] Can be used with Figure 2The performance of the components of the electronic device 100 shown in FIG. 1 corresponds to adding or omitting at least one component. In addition, it will be readily understood by those skilled in the art that the relative positions of the components may be changed to correspond to the performance or structure of the system.

[0145] on the other hand, Figure 2 The components shown in may be software components and / or hardware components, such as field programmable gate arrays (FPGAs) and application specific integrated circuits (ASICs).

[0146] In the following, reference will be made to Figures 4 to 7 An implementation example of the electronic device 100 is described.

[0147] Figure 4 is a control block diagram of an electronic device according to a second embodiment of the present disclosure, Figure 5 shows the appearance of a robot cleaner as an implementation example of the electronic device according to the second embodiment of the present disclosure, Figure 6 shows the bottom of a robot cleaner as an implementation example of the electronic device according to the second embodiment of the present disclosure, and Figure 7 An example of a state condition table for selecting a speech recognition model in an electronic device according to a second embodiment of an embodiment of the present disclosure is shown.

[0148] Reference Figure 4 , Figure 5 and Figure 6 , the electronic device 100 according to the second embodiment of the present disclosure may be implemented as a robot cleaner including a main body 101 and a buffer 102 positioned at the front side of the main body 101 .

[0149] like Figure 4 As shown in , the electronic device 100 according to the second embodiment of the present disclosure may include a microphone 110 , a transceiver 120 , a processor 130 , a memory 140 , and a driver 150 .

[0150] The above has been referred to Figure 2 The microphone 110 , the transceiver 120 , the processor 130 , and the memory 140 are described, and the driver 150 may include internal components of the robot cleaner. Hereinafter, features of the second embodiment different from the first embodiment will be described.

[0151] like Figure 5 As shown in , the body 101 of the electronic device 100 implemented as the robot cleaner according to the second embodiment of the present disclosure may be in a substantially cylindrical shape. More specifically, the body 101 may include a top surface 101a formed in a substantially circular shape, and a side surface 101b formed along the edge of the top surface 101a.

[0152] The plurality of microphones 110a and 110b may be positioned on the top surface 101a, but are not limited thereto. However, the microphone 110 may be positioned on the buffer 102, the front surface 101c, the side surface 101b, or the bottom (see Figure 6 ) on at least one of the above.

[0153] The buffer 102 can reduce the impact transmitted to the main body 101 when colliding with an obstacle, and sense the collision with the obstacle. The obstacle may be an object, a person, or an animal that interferes with the travel of the electronic device 100. The obstacle may include a wall separating the clean space, furniture positioned in the clean space, and a person or an animal located in the clean space.

[0154] like Figure 5 As shown in , when the electronic device 100 is traveling, the buffer 102 may face forward, and the direction facing the buffer 102 may be defined as the “front” direction of the electronic device 100 .

[0155] Components for performing the functions (movement and cleaning) of the robot cleaner may be provided inside and outside the main body 101 of the electronic device 100 , and the components may be included in the driver 150 .

[0156] like Figure 4 As shown in FIG. 1 , the driver 150 may include a motor 151 , a fan 152 , a brush 153 , and a wheel 154 .

[0157] The motor 151 may generate a driving force and transmit the generated driving force to at least one of the fan 152, the brush 153 or the wheel 154. To this end, a plurality of motors 151 may be provided to generate and transmit driving forces to the fan 152, the brush 153 and the wheel 154, respectively and independently.

[0158] The wheel 154 may include a first wheel 154 a positioned at the left side of the bottom of the electronic device 100 and a second wheel 154 b positioned at the right side of the bottom of the electronic device 100 , and is rotated by the motor 151 to move the body 101 .

[0159] When the main body 101 moves, the rotation axis of the roller 145 may rotate, and thus, the roller 145 may support the main body 101 without interfering with the travel of the electronic device 100 .

[0160] Furthermore, the electronic device 100 may include a driving circuit for supplying a driving current to the motor 151 , and a rotation sensor (eg, an encoder, a Hall sensor, etc.) for sensing the rotation of the wheel 154 .

[0161] like Figure 6As shown in , the brush 153 may be positioned in a suction port 101e formed at the bottom of the body 101. The brush 153 may rotate about a rotation axis positioned horizontally to the floor of the cleaning space to disperse dust collected on the floor to the inside of the suction port 101e.

[0162] Furthermore, the electronic device 100 may include various sensors for acquiring at least one of an operation mode, an operation intensity, an operation component, or an operation position.

[0163] When such various components operate in different operating states, characteristics of noise generated in the electronic device 100 may also change accordingly.

[0164] Therefore, the processor 130 can select a speech recognition model that can ensure a high recognition rate and high recognition performance according to the operating state of the electronic device 100, and perform speech recognition based on the selected speech recognition model. This operation is described above in the first embodiment, and the state condition for the operating state can be appropriately modified according to the implementation example of the electronic device 100.

[0165] Reference Figure 7 The state conditions may be predefined as conditions for an operation mode, an operation intensity, an operation component, and an operation position in an operation state, and a state condition table in which a speech recognition model is matched with an operation mode may be stored in the memory 140.

[0166] The first speech recognition model, the second speech recognition model, and the third speech recognition model stored in the memory 140 may be modeled to have different speech recognition parameters.

[0167] The operation modes may be classified according to the functions of the robot cleaner, and the operation intensity may be represented as at least one value of the rpm of the motor 151 , the torque of the motor 151 , the magnitude of the driving current provided to the motor 151 , the intensity of the fan 152 , or the rpm of the fan 152 .

[0168] When the operation mode of the electronic device 100 is the first mode, the rpm of the motor 151 is less than or equal to the first threshold value X1, and the operating component is the motor 151, the processor 130 may select a first speech recognition model from a plurality of speech recognition models.

[0169] When the operation mode of the electronic device 100 is the second mode, the rpm of the motor 151 is greater than the first threshold X1 and less than or equal to the second threshold X2, the operating components are the motor 151 and the fan 152, and the operating position is close to the wall, the processor 130 can select the second speech recognition model.

[0170] When the operation mode of the electronic device 100 is the third mode, the rpm of the motor 151 is greater than the third threshold value X3, and the operating components are the motor 151, the fan 152 and the brush 153, and the operating position is under the bed, the processor 130 can select the third voice recognition model.

[0171] In this way, the processor 130 can improve the speech recognition rate and speech recognition accuracy by selecting a speech recognition model suitable for a noise environment according to the detailed operation state of the electronic device 100 instead of unconditionally using a speech recognition model modeled in a normal environment.

[0172] In particular, when the speech recognition model is a wake-up word recognition model, the wake-up word can be accurately recognized despite the noise of the electronic device 100. Therefore, although the user no longer speaks the wake-up word, the wake-up word can be recognized quickly and accurately.

[0173] Like the first embodiment of the present disclosure, the processor 130 according to the second embodiment of the present disclosure can determine the sensitivity of the speech recognition model according to the operating state including at least one of the operating mode, operating intensity, operating component, or operating position, and change the speech recognition model or at least one of the sensitivity of the speech recognition model according to the change of the operating state. In addition, the processor 130 can change the sensitivity based on the user's intention, or generate a speech recognition model for the noise of the surrounding environment based on the operating state. This operation is described above in the first embodiment, and therefore, its further description will be omitted.

[0174] In the following, reference will be made to Figures 8 to 10 Another implementation example of the electronic device 100 is described.

[0175] Figure 8 is a control block diagram of an electronic device according to a third embodiment of the present disclosure. Fig. 9 shows the appearance of an air conditioner as an implementation example of the electronic device according to the third embodiment of the present disclosure, and Fig.10 is an exploded perspective view of an air conditioner as an implementation example of the electronic device according to the third embodiment of the present disclosure.

[0176] Reference Figure 8 , Fig. 9 and Fig.10 The electronic device 100 according to the third embodiment of the present disclosure can be implemented as an air conditioner, which includes a housing 10 having at least one outlet opening 41 (41a, 41b, 41c), a heat exchanger 20 for exchanging heat with air entering the interior of the housing 10, a blower 30 for circulating air to the inside or outside of the housing 10, and an outlet 40 for discharging the air blown from the blower 30 to the outside of the housing 10.

[0177] like Figure 8 As shown in , the electronic device 100 according to the third embodiment of the present disclosure may include a microphone 110 , a transceiver 120 , a processor 130 , a memory 140 , and a driver 160 .

[0178] The above has been referred to Figure 2 The microphone 110, the transceiver 120, the processor 130, and the memory 140 are described, and the driver 160 may include internal components of the air conditioner. Hereinafter, features of the third embodiment different from the first embodiment will be described.

[0179] like Fig. 9 As shown in , the housing 10 of the electronic device 100 implemented as the air conditioner according to the third embodiment of the present disclosure may include a front panel 10a in which at least one outlet opening 41 (41a, 41b, 41c) is formed, a rear panel 10b positioned behind the front panel 10a, a side panel 10c positioned between the front panel 10a and the rear panel 10b, and a top panel and a bottom panel 10d positioned above and below the side panel 10c. At least one outlet opening 41 may be circular, and at least two or more outlet openings 41 may be provided on the front panel 14 (or the rear panel 10b) in a manner spaced apart from each other. Fig. 9 10a). For example, the outlet opening 41 may include a first outlet opening 41a, a second outlet opening 41b, and a third outlet opening 41c. A plurality of outlet holes 42 may be provided in a portion of the front panel 10a.

[0180] In each of the front panel 10a and the side panel 10c, a first microphone 110d may be positioned, but is not limited thereto. However, the microphone 110 may be positioned in at least one of the top panel and the bottom panel 10d or the rear panel 10d. In addition, the microphone 110 may be positioned at an appropriate position for receiving the user's voice.

[0181] In the rear panel 10 b , a suction port 19 may be formed to suck external air into the interior of the housing 10 .

[0182] The suction port 19 may be positioned in the rear panel 10b behind the heat exchanger 20 to guide the outside air of the housing 10 into the inside of the housing 10. The air entering the inside of the housing 10 through the suction port 19 may pass through the heat exchanger 20 to absorb or dissipate heat. The air exchanged with heat by passing through the heat exchanger 20 may be discharged to the outside of the housing 10 via the outlet 40 by the blower 30.

[0183] The blower 30 may include a fan 162 and a grill 34 positioned in a protruding direction of the fan 162. According to some embodiments, the fan 162 may be positioned to correspond to at least one outlet opening 41, and the number of the fan 162 is not limited. For example, the fan 162 may include a first fan 162a, a second fan 162b, and a third fan 162c.

[0184] The blower 30 may include a motor 161 positioned at the center of the fan 162 to drive the fan 162. For example, the motor 161 may include a first fan motor 161a for driving the first fan 162a, a second fan motor 161b for driving the second fan 162b, and a third fan motor 161c for driving the third fan 162c. The motor 161 may rotate the fan 162, and the rotating fan 162 may generate an air flow passing through the heat exchanger 20.

[0185] The grille 34 may include a plurality of blades 35. By adjusting the number, shape, and installation angle of the plurality of blades 35, the direction or amount of air blown from the fan 162 to the outlet opening 41 may be adjusted.

[0186] At the center of the grille 34, a door actuator 163 may be positioned. The door actuator 163 and the motor 161 may be aligned in the front-rear direction. With this configuration, the plurality of blades 35 of the grille 34 may be positioned in front of the fan blades of the fan 162.

[0187] The blower 30 may include an air duct 36. The air duct 36 may be a circular shape surrounding the fan 162 to guide the air flow toward the fan 162.

[0188] The heat exchanger 20 may be positioned between the fan 162 and the suction port 19 to absorb heat from or transfer heat to the air entering through the suction port 19. The heat exchanger 20 may include a tube 21 and a header 22 coupled to the upper and lower ends of the tube 21. However, the type of the heat exchanger 20 is not limited.

[0189] Components for performing functions of air conditioning such as blowing, temperature adjustment, air purification, etc. may be provided inside and outside the housing 10 of the electronic device 100 , and these components may be included in the driver 160 .

[0190] The outlet opening 41 may be opened and closed by a door 60, and may include a door blade 62 (62a, 62b, or 62c). The door actuator 163 may operate the door blade 62 to open or close the door 60. The door actuator 163 may space the door blade 62 from the end 43 of the discharge guide 45 to open the door 60, and may contact the door blade 62 with the end 43 of the discharge guide 45 to close the door 60.

[0191] The discharge guide 45 may include a guide body 46 and a guide hole 47. The guide body 46 may form a first discharge path therein. The guide body 46 may be in the shape of a cylinder having a hollow interior. More specifically, the guide body 46 may be in the shape of a duct having one end facing the blower unit 30 and the other end facing the outlet opening 41. The guide hole 47 may allow the second discharge path to pass therethrough. The guide hole 47 may be formed in the guide body 46. The shape of the guide hole 47 may not be limited, and the guide hole 47 may have any structure that may be formed in the guide body 46 and enables air to flow in the outer direction of the guide body 46. In the current embodiment, the guide hole 47 may be a plurality of holes formed along the circumference of the guide body 46. The door blade 62 may include a blade body 63 and a blade coupling portion 64, wherein the blade body 63 is circular to correspond to the outlet opening 41, and the blade coupling portion 64 extends from the blade body 63 and is connected to the door actuator 163.

[0192] The driver 160 may include a motor 161 , a fan 162 , and a door actuator 163 .

[0193] In addition, the electronic device 100 may include a drive circuit for supplying a drive current to the motor 161 and a rotation sensor (e.g., an encoder, a Hall sensor, etc.) for sensing the rotation of the fan 162, and may include various sensors for obtaining at least one of an operating mode, an operating intensity, an operating component, or an operating position.

[0194] When such various components operate in different operating states, characteristics of noise generated in the electronic device 100 may also change accordingly.

[0195] Therefore, the processor 130 can determine the sensitivity of the speech recognition model that can ensure a high recognition rate and high recognition performance according to the operating state of the electronic device 100, and perform speech recognition based on the determined sensitivity. This operation has been described above in the first embodiment, and the detailed conditions for the operating state according to the sensitivity can be appropriately modified according to the implementation example of the electronic device 100.

[0196] For example, the condition for the operating state according to the sensitivity can be set as the condition for the operating mode, and the operating mode of the air conditioner can be classified into a first mode and a second mode, wherein the first mode is a breeze mode and the second mode is a strong wind mode. In this case, the processor 130 can determine the sensitivity of the speech recognition model as a first level in the first mode, and can determine the sensitivity of the speech recognition model as a second level having a higher sensitivity than the first level in the second mode.

[0197] In addition, the condition for the operating state according to the sensitivity can be set as a condition for at least one of the operating mode, operating intensity, operating component, or operating position of the air conditioner. The processor 130 can identify the state condition that satisfies the operating state among the predefined conditions for the operating state according to the sensitivity, and determine the sensitivity corresponding to the state condition that satisfies the operating state as the sensitivity of the speech recognition model.

[0198] To this end, a sensitivity table in which the sensitivity of the speech recognition model is matched with the operating state may be stored in the memory 140. The processor 130 may compare the operating state of the electronic device 100 with the sensitivity table to determine the sensitivity of the speech recognition model corresponding to the operating state.

[0199] In this way, by determining the sensitivity of the speech recognition model according to the operation state, the processor 130 can reduce the influence generated by the noise of the electronic device 100. Therefore, the accuracy and efficiency of speech recognition can be improved.

[0200] In particular, when the speech recognition model is a wake-up word recognition model, the wake-up word can be accurately recognized through sensitivity adjustment despite the noise of the electronic device 100. Therefore, although the user no longer speaks the wake-up word, the wake-up word can be recognized quickly and accurately.

[0201] Like the first embodiment of the present disclosure, when the operating state changes, the processor 130 according to the third embodiment of the present disclosure can change the sensitivity based on the changed operating state, and recognize the user's voice based on the changed sensitivity. To this end, the processor 30 can acquire the operating state in real time or at predefined time intervals. In addition, in addition to the operating state, the processor 130 can change the sensitivity of the voice recognition model according to the user's intention, and generate a voice recognition model for the noise of the surrounding environment based on the operating state. This operation is described above in the first embodiment, and therefore, further description thereof will be omitted.

[0202] At least one component may be added or omitted to correspond to Figure 4 and Figure 8 1. In addition, one of ordinary skill in the art will readily appreciate that the relative positions of the components may be changed to correspond to the performance or structure of the system.

[0203] on the other hand, Figure 4 and Figure 8 The components shown in may be software components and / or hardware components, such as FPGAs and ASICs.

[0204] Fig.11 is a flowchart illustrating a voice recognition method of an electronic device according to an embodiment of the present disclosure.

[0205] Reference Fig.11 In operation 901, the electronic device 100 according to an embodiment of the present disclosure may determine whether to acquire an operating state. Herein, the operating state of the electronic device 100 may be a state factor that may affect noise generated in the electronic device 100, that is, may be a state factor from which the noise of the electronic device 100 is estimated. The operating state may include at least one of an operating mode, an operating intensity, an operating component, or an operating position of the electronic device 100.

[0206] After acquiring the operation state (“YES” in operation 901 ), in operation 902 , the electronic device 100 may select a speech recognition model based on the operation state.

[0207] More specifically, the electronic device 100 may select a speech recognition model from a plurality of speech recognition models pre-stored in the memory 140 or a server (not shown) based on the operation state.

[0208] The electronic device 100 can identify a state condition that satisfies an operating state from among predefined state conditions for a plurality of speech recognition models, and select a speech recognition model corresponding to the state condition that satisfies the operating state as the speech recognition model corresponding to the operating state. At this time, the state condition for each of the plurality of speech recognition models can be defined as a condition for the operating state according to design or according to user input.

[0209] To this end, a state condition table in which state conditions for operating states are matched with speech recognition models may be stored in the memory 140. The electronic device 100 may compare the acquired operating state with the state condition table to determine a speech recognition model corresponding to the operating state.

[0210] In operation 903 , by selecting a speech recognition model corresponding to the acquired operation state from among the speech recognition models for the operation state, the electronic device 100 may recognize the user's speech based on the selected speech recognition model.

[0211] In this way, by selecting a speech recognition model suitable for a noisy environment according to the operating state of the electronic device 100, rather than unconditionally using a speech recognition model modeled in a normal environment, and using the speech recognition model for speech recognition, the speech recognition rate and speech recognition accuracy can be improved.

[0212] In particular, when the speech recognition model is a wake-up word recognition model, the wake-up word can be accurately recognized despite the noise of the electronic device 100. Therefore, although the user no longer speaks the wake-up word, the wake-up word can be recognized quickly and accurately.

[0213] Fig.12is a flowchart illustrating a voice recognition method of an electronic device according to an embodiment of the present disclosure.

[0214] Reference Fig.12 According to an embodiment of the present disclosure, the electronic device 100 may acquire an operation state, and after acquiring the operation state ("Yes" in operation 910), in operation 911, the electronic device 100 may select a speech recognition model based on the operation state. Operations 910 and 911 may be related to Fig.11 Operation 901 is the same as operation 902.

[0215] Then, the electronic device 100 may determine whether the operation state is changed in operation 912. The electronic device 100 may check the operation state in real time or at predefined time intervals.

[0216] When the operation state changes ("Yes" in operation 912), the electronic device 100 may change the speech recognition model to a speech recognition model corresponding to the changed operation state in operation 913. That is, the electronic device 100 may reselect the speech recognition model based on the changed operation state.

[0217] In operation 914 , the electronic device 100 may recognize the user's voice based on the changed voice recognition model.

[0218] Therefore, since the speech recognition model is changed according to the operation state of the electronic device 100, a speech recognition model suitable for a noise environment can be used according to the operation state of the electronic device 100. Therefore, the accuracy and efficiency of speech recognition can be improved.

[0219] Fig.13 is a flowchart illustrating a voice recognition method of an electronic device according to an embodiment of the present disclosure.

[0220] Reference Fig.13 According to an embodiment of the present disclosure, the electronic device 100 may acquire an operation state, and after the electronic device 100 acquires the operation state ("Yes" in operation 921), in operation 922, the electronic device 100 may select a speech recognition model based on the operation state. Operations 921 and 922 may be related to Fig.11 Operation 901 is the same as operation 902.

[0221] After the electronic device 100 selects a speech recognition model, in operation 923, the electronic device 100 may determine the sensitivity of the selected speech recognition model based on the operating state. Sensitivity may be an example of a parameter of the speech recognition model. In the current embodiment of the present disclosure, sensitivity may be set as a parameter of the speech recognition model, however, the electronic device 100 may determine various speech recognition related parameters based on the operating state.

[0222] More specifically, the electronic device 100 can identify the state condition that satisfies the operating state from the predefined conditions for the operating state for the sensitivity, and determine the sensitivity corresponding to the state condition that satisfies the operating state as the sensitivity of the speech recognition model. To this end, a sensitivity table in which the sensitivity of the speech recognition model is matched with the operating state can be stored in the memory 140. The electronic device 100 can compare the operating state of the electronic device 100 with the sensitivity table to determine the sensitivity of the speech recognition model corresponding to the operating state.

[0223] On the other hand, Fig.13 , it is shown that operation 923 is performed after operation 922 is performed, however, operation 922 and operation 923 may be performed at the same time.

[0224] In operation 924, the electronic device 100 may determine whether the operation state is changed. To this end, the electronic device 100 may check the operation state in real time or at predefined time intervals.

[0225] When the electronic device 100 determines that the operating state has changed ("Yes" in operation 924), the electronic device 100 may change the sensitivity based on the changed operating state in operation 925. More specifically, the electronic device 100 may determine the sensitivity again based on the changed operating state, thereby changing the sensitivity of the speech recognition model to the sensitivity corresponding to the changed operating state.

[0226] Thereafter, in operation 926 , the electronic device 100 may recognize the user's voice based on the changed sensitivity.

[0227] Thus, since the sensitivity of the speech recognition model is determined according to the operation state and the sensitivity is changed according to the change of the operation state, the influence caused by the noise of the electronic device 100 can be reduced. Therefore, the accuracy and efficiency of speech recognition can be improved.

[0228] In particular, when the speech recognition model is a wake-up word recognition model, the wake-up word can be accurately recognized through sensitivity adjustment despite the noise of the electronic device 100. Therefore, although the user no longer speaks the wake-up word, the wake-up word can be recognized quickly and accurately.

[0229] On the other hand, the disclosed embodiments may be implemented in the form of a recording medium storing instructions executable by a computer. The instructions may be stored in the form of program code, and when executed by a processor, the instructions may create a program module to perform the operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

[0230] The computer-readable recording medium may include all kinds of recording media storing instructions that can be interpreted by a computer. For example, the computer-readable recording medium may be a ROM, RAM, a magnetic tape, a magnetic disk, a flash memory, an optical data storage device, and the like.

[0231] While the present disclosure has been shown and described with reference to various embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made without departing from the spirit and scope of the present disclosure as defined by the appended claims and their equivalents.

Claims

1. An electronic device, comprising: a microphone configured to receive a user's voice; A memory, used for storing a plurality of speech recognition models; as well as At least one processor configured to: selecting a speech recognition model from the plurality of speech recognition models stored in the memory based on an operating state of the electronic device, determining a sensitivity of the selected speech recognition model based on the operating state, determining the user's intention based on morpheme analysis or speech behavior analysis of the user's speech, changing the sensitivity based on the user intent, and The user's voice received by the microphone is recognized based on the selected voice recognition model and the sensitivity.

2. The electronic device according to claim 1, wherein: The operation state includes at least one of an operation mode or an operation intensity of the electronic device.

3. The electronic device according to claim 1, wherein: The at least one processor is further configured to select a speech recognition model corresponding to a state condition satisfying the operating state among a plurality of state conditions as a speech recognition model corresponding to the operating state, wherein the plurality of state conditions are respectively matched with the plurality of speech recognition models.

4. The electronic device according to claim 1, wherein: The at least one processor is further configured to: When the operating state changes, changing the selected speech recognition model based on the changed operating state, and The user's voice is recognized based on the changed operation state.

5. The electronic device according to claim 1, wherein: The at least one processor is further configured to, when the operating state changes, change the sensitivity based on the changed operating state.

6. The electronic device according to claim 1, wherein: The at least one processor is further configured to: determining a first pattern for ambient noise based on the operating state, and A speech recognition model is generated based on the determined first pattern.

7. The electronic device according to claim 6, wherein: The at least one processor is further configured to: determining a second pattern of noise for the electronic device based on the operating state, and The first pattern for noise of the surrounding environment is determined based on the second pattern.

8. The electronic device according to claim 1, wherein: The multiple speech recognition models include multiple speech recognition models for recognizing wake-up words.

9. The electronic device according to claim 1, wherein: The electronic device includes at least one of a cleaner, an air conditioner, a refrigerator, a washing machine, or a clothing care appliance.

10. An electronic device comprising: a microphone configured to receive a user's voice; a transceiver configured to communicate with a server; as well as At least one processor configured to: selecting a speech recognition model among a plurality of speech recognition models received from the server based on an operating state of the electronic device, determining a sensitivity of the selected speech recognition model based on the operating state, determining the user's intention based on morpheme analysis or speech behavior analysis of the user's speech, changing the sensitivity based on the user intent, and The user's voice received by the microphone is recognized based on the selected voice recognition model and the sensitivity.

11. A speech recognition method for an electronic device, the speech recognition method comprising: Receive user's voice; selecting a speech recognition model from a plurality of pre-stored speech recognition models based on an operating state of the electronic device; determining a sensitivity of the selected speech recognition model based on the operating state, determining the user's intention based on morpheme analysis or speech behavior analysis of the user's speech, changing the sensitivity based on the user intent, and The user's voice is recognized based on the selected voice recognition model and the sensitivity.

12. The speech recognition method according to claim 11, wherein: Selecting the speech recognition model includes: A speech recognition model corresponding to a state condition satisfying the operating state among a plurality of state conditions is selected as a speech recognition model corresponding to the operating state, wherein the plurality of state conditions are matched with the plurality of speech recognition models respectively.

13. The speech recognition method according to claim 11, further comprising: When the operating state changes, the selected speech recognition model is changed based on the changed operating state, Herein, recognizing the user's voice includes recognizing the user's voice based on the changed operation state.

14. The speech recognition method according to claim 11, further comprising: When the operating state changes, the sensitivity of the selected speech recognition model is changed based on the changed operating state, Wherein, recognizing the user's voice based on the selected voice recognition model and the sensitivity includes recognizing the user's voice based on the selected voice recognition model and the changed sensitivity.

15. The speech recognition method according to claim 11, further comprising: determining a first pattern for ambient noise based on the operating state; as well as A speech recognition model is generated based on the determined first pattern.

16. The speech recognition method according to claim 15, further comprising: determining a second pattern of noise for the electronic device based on the operating state, The determining of the first mode includes determining the first mode for the noise of the surrounding environment based on the second mode.

Images