Speech information recognition method, system, device and storage medium

Abstract

This application discloses a method, system, device, and storage medium for recognizing voice information. The method selects a suitable smart terminal for command matching by determining whether the voice information contains a wake-up word. Furthermore, this method distinguishes voice command words related to vehicle system operation according to user frequency, distributing and storing frequently used and infrequently used command words. Frequently used command words are stored in the local command library of the vehicle terminal, while infrequently used command words are stored in a cloud command library. This reduces the performance consumption of the vehicle terminal and effectively avoids accidental wake-up of the vehicle system, greatly improving the user experience.

Description

Technical Field

[0001] This application relates to the field of vehicle networking, and in particular to a method, system, device and storage medium for recognizing voice information. Background Technology

[0002] Currently, to enhance driver safety and convenience when using in-vehicle devices, some systems have incorporated voice recognition. By using voice wake-up commands to interact with these devices, drivers can avoid manual control, thus reducing distractions and minimizing safety hazards while driving. For example, in automotive intelligent control systems, corresponding voice wake-up commands can be used to control features like the air conditioning, or voice commands can be used to coordinate in-vehicle and mobile terminals, enabling the in-vehicle terminal to share resources with the mobile terminal.

[0003] In existing technologies, in-vehicle terminals store all user commands related to vehicle system operation. Current voice recognition technology stores all commonly used commands and searches its local command library for the corresponding command based on the user's voice command, executing the desired operation. However, due to increasingly complex user needs and performance differences among in-vehicle terminals, the ever-increasing number of commands poses a significant challenge to terminal performance. Furthermore, because the in-vehicle terminal's local command library stores a large number of commands, erroneous responses are highly likely during user interaction (e.g., a user says "turn up the volume" while talking to someone else, but the command is not directed to the vehicle system; the system will still respond to the command and adjust the volume). This method of using voice commands wastes vehicle system resources and degrades the user experience. Therefore, a more efficient method for recognizing voice information is needed. Summary of the Invention

[0004] One objective of this application is to provide a method, system, device, and storage medium for recognizing voice information. Its advantage lies in that when recognizing a user's voice information, it determines whether the voice information contains a wake-up word and selects an appropriate instruction library for instruction matching, effectively avoiding the situation where the vehicle system is mistakenly woken up, and greatly improving the user experience.

[0005] Another objective of this application is to provide a voice information recognition method, system, device, and storage medium. Its advantage lies in distinguishing hot words related to vehicle operation according to the user's usage frequency, distributing and storing frequently used and infrequently used command words, storing frequently used command words in the local command library of the vehicle terminal, and storing infrequently used command words in the cloud command library. This not only reduces the performance consumption of the vehicle terminal, but also effectively avoids the situation of the vehicle system being accidentally woken up, greatly improving the user experience.

[0006] To achieve the aforementioned objective, in a first aspect, embodiments of this application provide a method for recognizing voice information, the method comprising the following steps: acquiring a user's voice information; selecting different methods to acquire user needs based on whether the voice information contains a wake word; acquiring operation instructions based on the user needs; and performing corresponding operations based on the operation instructions.

[0007] In this method, a specific recognition terminal is selected for the voice information by determining whether the voice information contains a wake word. The hot words corresponding to the voice information are no longer extracted from the hot word library of the vehicle terminal, but the appropriate recognition terminal is selected for the voice information based on whether the voice information contains a wake word, which avoids false responses and improves the recognition efficiency of voice information.

[0008] In an optional implementation of the first aspect, the method further includes: periodically detecting the usage frequency of hot words in the local hot word library and the cloud hot word library; if the usage frequency of a local hot word in the local hot word library within a target period is not greater than a target threshold, deleting the local hot word from the local hot word library and storing the hot word in the cloud hot word library, wherein the hot words in the cloud hot word library include hot words whose user usage frequency is not greater than the target threshold; if the usage frequency of a cloud hot word in the cloud hot word library within a target period is greater than the target threshold, storing the cloud hot word in the local hot word library, wherein any hot word in the local hot word library is a hot word whose user usage frequency is greater than the target threshold.

[0009] It should be understood that the hot words stored in the local hot word library are those used by users more frequently than a target threshold, while the hot words stored in the cloud hot word library include not only those used more frequently than the target threshold but also those used less frequently than the target threshold. Therefore, while ensuring smooth response to high-frequency hot words, the possibility of false responses to low-frequency hot words is reduced. Furthermore, different users may have different usage preferences at different times. For example, sometimes users may frequently need to make calls while driving, while other times they may need to frequently adjust the vehicle's hardware. Therefore, in this embodiment, by periodically detecting the usage frequency of hot words and promptly using the hot words from the local and cloud hot word libraries, the continuous high efficiency of user voice information recognition can be ensured.

[0010] Secondly, embodiments of this application provide a voice information recognition system, the system comprising: an in-vehicle terminal and a cloud, wherein the in-vehicle terminal and the cloud are communicatively connected; the in-vehicle terminal acquires user voice information; determines whether the voice information contains a wake-up word; if the voice information does not contain a wake-up word, the in-vehicle terminal acquires user needs by recognizing local hot words matching the voice information, and acquires operation instructions corresponding to the user needs; if the voice information contains a wake-up word, the in-vehicle terminal sends the voice information to the cloud; the cloud receives the voice information, acquires user needs by recognizing cloud hot words matching the voice information, and sends operation instructions corresponding to the user needs to the in-vehicle terminal; the in-vehicle terminal executes corresponding operations according to the acquired operation instructions.

[0011] In one possible implementation of the second aspect, if the voice information contains a wake-up word, the vehicle terminal obtains user needs by recognizing local hot words that match the voice information, and obtains operation instructions corresponding to the user needs.

[0012] It should be understood that when the storage and recognition performance of the vehicle-mounted terminal exceeds a certain level, the vehicle-mounted terminal can independently complete the storage of hot words for voice information recognition. Therefore, in this embodiment, the storage and recognition performance of the vehicle-mounted terminal is sufficient to independently complete the process of storing hot words for voice information recognition. The system no longer needs to work in conjunction with the cloud, thus reducing the data forwarding process and further improving the voice information recognition efficiency.

[0013] Thirdly, this application provides another voice information recognition system, the system comprising: an in-vehicle terminal, a cloud, and a smart terminal, wherein the in-vehicle terminal, the cloud, and the smart terminal are communicatively connected; the in-vehicle terminal acquires user voice information; determines whether the voice information contains a wake-up word; if the voice information does not contain a wake-up word, the in-vehicle terminal acquires user needs by recognizing local hot words matching the voice information, acquires operation instructions corresponding to the user needs, and sends the operation instructions corresponding to the user needs to the smart terminal; if the voice information contains a wake-up word, the in-vehicle terminal sends the voice information to the smart terminal; the smart terminal sends the voice information to the cloud; the cloud receives the voice information, acquires user needs by recognizing cloud hot words matching the voice information, and sends operation instructions corresponding to the user needs to the smart terminal; the smart terminal executes corresponding operations according to the acquired operation instructions.

[0014] In one possible implementation of the third aspect, if the voice information contains a wake-up word, the vehicle terminal obtains user needs by recognizing local hot words that match the voice information, obtains operation instructions corresponding to the user needs, and sends the operation instructions to the smart terminal.

[0015] Alternatively, if the voice information contains a wake-up word, the vehicle terminal sends the voice information to the smart terminal; the smart terminal obtains user needs by recognizing local hot words of the smart terminal that match the voice information, and obtains operation instructions corresponding to the user needs.

[0016] It should be understood that when the storage and recognition performance of the vehicle-mounted terminal and the smart terminal are above a certain level, both the vehicle-mounted terminal and the smart terminal can independently complete the storage of hot words for voice information recognition. Therefore, in this embodiment, the storage and recognition performance of the vehicle-mounted terminal and the smart terminal are sufficient to independently complete the process of storing hot words for voice information recognition. The system no longer needs to work collaboratively with the cloud, thus reducing the data forwarding process and further improving the efficiency of voice information recognition.

[0017] Fourthly, embodiments of this application provide an electronic device, including: a memory for storing a program; and a processor for executing the program stored in the memory. When the program is executed, the processor is configured to perform the steps of the method as described in the first aspect and any of the optional implementations.

[0018] Fifthly, embodiments of this application provide a computer-readable storage medium, characterized in that the computer storage medium stores a computer program, the computer program including program instructions, which, when executed by a processor, cause the processor to perform the steps of the method as described in the first aspect and any optional implementation. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments or background art of this application, the accompanying drawings used in the embodiments or background art of this application will be briefly introduced below.

[0020] Figure 1 A flowchart illustrating a method for recognizing voice information provided in an embodiment of this application;

[0021] Figure 2 A schematic diagram illustrating the hot word registration and response process provided in this application embodiment;

[0022] Figure 3 An interactive flowchart of a voice information recognition method provided in an embodiment of this application;

[0023] Figure 4 This is a schematic diagram of the structure of a voice information recognition system provided in an embodiment of this application;

[0024] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described below in conjunction with the accompanying drawings.

[0026] The terms "first" and "second," etc., used in the specification, claims, and drawings of this application are used only to distinguish different objects and not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0027] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0028] In this application, "at least one (item)" means one or more, "more than one" means two or more, "at least two (items)" means two or three or more, and "and / or" is used to describe the relationship between related objects, indicating that there can be three relationships. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the related objects before and after are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items. For example, at least one (item) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c".

[0029] This invention provides a method and related apparatus for recognizing voice information. To more clearly describe the solution of this invention, some knowledge related to the method and related apparatus for recognizing voice information provided in this application will be introduced below.

[0030] Hot word registration: Hot words refer to the command words that users use to control smart devices by voice during voice interaction. Hot word registration refers to storing these hot words in a unified command library and setting corresponding matching rules for the hot words in the command library and the text information converted from user voice information.

[0031] Wake word: For most voice interaction products on the market, a glance or a gesture is not enough to get their attention. Therefore, it is necessary to define a word to switch the product from standby mode to working mode, which is called "wake word".

[0032] In today's technology, scenario-based hot words are registered synchronously in the vehicle's cloud system. Due to the limitations of vehicle system performance, there is an upper limit to the number of hot words. If a large number of dialogues are registered on the vehicle system, the performance consumption of the vehicle system when processing dialogue recognition will increase significantly. At the same time, because there are many hot words registered on the vehicle system, unintentional words spoken by users in the car may accidentally wake up the vehicle system and execute corresponding operations, affecting the user experience.

[0033] To address existing problems in current technologies, this application provides a voice information recognition method. This method selects a suitable smart terminal for command matching by determining whether the voice information contains a wake-up word. Furthermore, this method differentiates voice commands related to vehicle system operation according to user frequency, distributing and storing frequently used and infrequently used commands. Frequently used commands are stored in the vehicle terminal's local command library, while infrequently used commands are stored in a cloud-based command library. This method reduces the performance consumption of the vehicle terminal and effectively prevents accidental wake-up of the vehicle system, greatly improving the user experience. Please refer to [link / reference] for details. Figure 1 .

[0034] Figure 1 This is a flowchart illustrating a method for recognizing voice information provided in an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0035] 101. Obtain the user's voice information.

[0036] The vehicle-mounted terminal monitors the surrounding environment for real time. When it detects voice information emitted by a user in the environment, it acquires that user's voice information. The vehicle-mounted terminal can be a vehicle, vehicle-mounted equipment (such as an On-Board Unit (OBU)), a wireless terminal in self-driving technology, a terminal in transportation safety, etc. Any device that can communicate with roadside equipment, vehicles, or vehicle management platforms falls within the protection scope of the vehicle-mounted terminal.

[0037] 102. Depending on whether the voice information contains a wake word, select different methods to obtain user needs.

[0038] After the vehicle terminal receives the voice information, it will determine whether there is a wake-up word in the voice information.

[0039] If the voice message does not contain a wake word, the vehicle terminal will obtain the user's request.

[0040] When the voice information contains a wake-up word, the vehicle terminal sends the voice information to the smart terminal, which then forwards it to the cloud. The cloud then recognizes the voice information and obtains the user's request. The smart terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The cloud can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The smart terminal and the cloud can be directly or indirectly connected via wired or wireless communication, which is not limited herein. It should be understood that in some embodiments, the voice information issued by the user begins with a specific wake-up word, such as the specific wake-up word "Hi, Siri" on Apple phones and "Xiaoyi Xiaoyi" on Huawei phones. When the vehicle terminal hears the wake-up word, the vehicle terminal begins to receive subsequent voice information from the user.

[0041] 103. Obtain operation instructions based on the user's requirements.

[0042] When the voice information does not contain a wake-up word, after receiving the voice information, the vehicle terminal converts the vocabulary in the voice information into computer-readable input, such as text content. After recognizing the voice information and obtaining the recognition result (i.e., the text content), the vehicle terminal searches for hot words matching the text content from a local hot word database to obtain the operation command.

[0043] When the voice information contains a wake-up word, after receiving the voice information, the vehicle terminal sends the voice information to the smart terminal, which then forwards it to the cloud. The cloud converts the vocabulary in the voice information into computer-readable input, such as text. After recognizing the voice information and obtaining the recognition result (i.e., the text content), the cloud searches for matching hot words in its cloud hot word database to obtain the operation command.

[0044] In some embodiments, when the voice information contains a wake-up word, after receiving the voice information, the vehicle terminal converts the vocabulary content in the voice information into computer-readable input, such as text content, and sends the text content to the smart terminal, which then forwards it to the cloud. The cloud can directly use the text content to search for hot words that match the text content from cloud hot words to obtain the operation command.

[0045] 104. Perform the corresponding operation according to the operation instructions.

[0046] It should be understood that the execution end of the operation command may be the in-vehicle terminal (e.g., the operation command is an instruction to control vehicle-related hardware devices) or the smart terminal (e.g., the operation command is an instruction to open an app installed on a mobile phone). Therefore, regardless of whether the operation command is acquired by the in-vehicle terminal or the cloud, the execution end of the operation command must be determined.

[0047] When the operation command is acquired by the vehicle-mounted terminal (i.e., when the voice information does not contain a wake-up word), the vehicle-mounted terminal will determine the execution end of the operation command. If the execution end of the operation command is the vehicle-mounted terminal, the vehicle-mounted terminal will directly execute the operation command; if the execution end of the operation command is the smart terminal, the vehicle-mounted terminal will forward the operation command to the smart terminal, and the smart terminal will execute the operation command upon receiving it.

[0048] When the operation command is obtained from the cloud (i.e., when the voice information contains a wake-up word), the cloud sends the operation command to the smart terminal, and the smart terminal determines the execution end of the operation command. When the execution end of the operation command is the smart terminal, the smart terminal directly executes the operation command after receiving it; when the execution end of the operation command is the vehicle terminal, the smart terminal forwards the operation command to the vehicle terminal, and the vehicle terminal executes the operation command after receiving it.

[0049] In one optional implementation, the vehicle-mounted terminal and the cloud are directly communicatively connected. When the voice information contains a wake-up word, after receiving the voice information, the vehicle-mounted terminal sends the voice information to the cloud. The cloud then converts the vocabulary in the voice information into computer-readable input, such as text content. After recognizing the voice information and obtaining the recognition result (i.e., the text content), the cloud searches for hot words matching the text content in its cloud hot word database to obtain the operation instruction, and sends the operation instruction to the vehicle-mounted terminal. The vehicle-mounted terminal then completes the corresponding operation according to the operation instruction.

[0050] It should be understood that, in this embodiment, the hot words in the local hot word library of the in-vehicle terminal are all hot words whose frequency of use by the user is greater than a target threshold. The hot words stored in the cloud hot word library include not only hot words whose frequency of use by the user is greater than the target threshold, but also hot words whose frequency of use by the user is not greater than the target threshold. Therefore, while ensuring that high-frequency hot words can be responded to smoothly, the possibility of low-frequency hot words being responded to incorrectly is reduced. In addition, different users may have different usage preferences at different times. For example, sometimes users may frequently need to make calls in the car while driving, while at other times users may frequently need to adjust the hardware devices in the car. Therefore, preferably, in some embodiments, the vehicle terminal and the cloud will periodically detect the usage frequency of the hot words stored therein within a certain period. If the usage frequency of a certain local hot word in the local hot word library within the target period is not greater than the target threshold, then the hot word can be considered as a low-frequency hot word that the user has used less recently, and the local hot word will be deleted from the local hot word library and stored in the cloud hot word library; if the usage frequency of a certain hot word in the cloud hot word library within the target period is greater than the target threshold, then the hot word can be considered as a high-frequency hot word that the user has used more frequently recently, and the hot word will be stored in the local hot word library.

[0051] To further illustrate the registration and response methods for hot words in the local hot word library and the cloud-based hot word library, this application embodiment provides a schematic diagram of the hot word registration and response process. Please refer to the following for details. Figure 2 .

[0052] Figure 2 This application provides a schematic diagram of a hot word registration and response process, as shown in the embodiments. Figure 2 As shown in this embodiment, during hot word registration, hot words frequently used by users will be categorized based on data from vehicle network industry statistics into hot words whose frequency of use exceeds a preset target threshold (i.e., ...). Figure 2High-frequency hot words) and hot words whose user usage frequency is lower than the aforementioned preset target threshold (i.e., low-frequency hot words). Figure 2 (Not shown in the text). High-frequency hot words are registered in a local hot word database within the vehicle terminal, which can be the aforementioned... Figure 1 The description mentions in-vehicle terminals; low-frequency hot keywords are registered in the cloud, and the cloud can be the aforementioned... Figure 2 The cloud mentioned in the description means that, at the time of manufacture, the vehicle terminal and the cloud have already been distributed and stored according to the frequency of use of hot words.

[0053] In some embodiments, the hot words stored in the cloud-based hot word library include not only hot words whose frequency of use is below a target threshold (hereinafter referred to as low-frequency hot words), but also hot words whose frequency of use is above the target threshold (i.e., high-frequency hot words). That is, when a user cannot determine whether the hot word corresponding to their expected operation is a low-frequency hot word or a high-frequency hot word, they can directly speak voice information containing the wake word and the hot word, so that the voice information can be successfully responded to.

[0054] It should be understood that different users have different usage preferences and habits, and the same user may also have different usage preferences and habits at different times. For example, sometimes a user may frequently need to make calls while driving, while at other times a user may frequently need to adjust the hardware devices in the car. Therefore, the in-vehicle terminal and the cloud will periodically detect the usage frequency of the hot words they store within a certain period. If the usage frequency of a local hot word in the local hot word library is not greater than the target threshold within the target period, then the hot word can be considered a low-frequency hot word that the user has used less recently, and the hot word will be deleted from the local hot word library and stored in the cloud hot word library; if the usage frequency of a hot word in the cloud hot word library is greater than the target threshold within the target period, then the hot word can be considered a high-frequency hot word that the user has used more frequently recently, and the hot word will be stored in the local hot word library.

[0055] In actual use, after the in-vehicle terminal obtains the user's voice information, it will determine whether the voice information contains a wake-up word. When the voice information does not contain a wake-up word, the in-vehicle terminal will respond to the voice information, retrieve a local hot word matching the voice information from the local hot word library, and obtain an operation command corresponding to the local hot word; when the voice information contains a wake-up word, the in-vehicle terminal will send the voice information to the cloud for response, retrieve a cloud hot word matching the voice information from the cloud hot word library, and obtain an operation command corresponding to the cloud hot word.

[0056] After receiving the operation instruction, the next step is to determine the execution device of the cloud-based hot words or the local hot words (i.e., Figure 2 The intelligent terminal (either the vehicle-mounted terminal or the intelligent terminal itself) handles the response logic for hot words.

[0057] It should be understood that the operation instructions contained in the hot words may require execution by either the in-vehicle terminal or the smart terminal. When the operation instruction is obtained by the in-vehicle terminal (i.e., when the voice information does not contain a wake-up word), the in-vehicle terminal will determine the execution end of the operation instruction. If the execution end of the operation instruction is the in-vehicle terminal (in which case the smart terminal is the in-vehicle terminal), the in-vehicle terminal will directly execute the operation instruction; if the execution end of the operation instruction is the smart terminal (in which case the smart terminal is the smart terminal), the in-vehicle terminal will forward the operation instruction to the smart terminal, and the smart terminal will execute the operation instruction upon receiving it.

[0058] When the operation command is obtained from the cloud (i.e., when the voice information contains a wake-up word), the cloud sends the operation command to the smart terminal, and the smart terminal determines the execution end of the operation command. If the execution end of the operation command is the smart terminal (in this case, the smart terminal is the smart terminal), the smart terminal directly executes the operation command after receiving it; if the execution end of the operation command is the vehicle terminal (in this case, the smart terminal is the vehicle terminal), the smart terminal forwards the operation command to the vehicle terminal, and the vehicle terminal executes the operation command after receiving it. (This is in conjunction with...) Figure 2 The description of the speech information recognition method shown in this application provides an interactive flowchart of a speech information recognition method. Please refer to... Figure 3 .

[0059] Figure 3 This application provides an interactive flowchart of a voice information recognition method, which involves an in-vehicle terminal, a smart terminal, and a cloud. Specifically, the in-vehicle terminal, smart terminal, and cloud can be the aforementioned... Figure 1 and Figure 2 The description refers to the in-vehicle terminal, smart terminal, and cloud. The method includes, but is not limited to, the following steps:

[0060] 301. Commonly used keywords for vehicle terminal registration.

[0061] When the vehicle terminal leaves the factory, hot words that are used more frequently than a preset threshold are stored in the local hot word library of the vehicle terminal, thus completing the hot word registration of the vehicle terminal.

[0062] 302. Register all hot keywords in the cloud.

[0063] All hot words whose frequency of use by users exceeds the preset threshold and hot words whose frequency of use by users does not exceed the preset threshold are stored in the cloud hot word library in the cloud, thus completing the cloud hot word registration.

[0064] 303. The vehicle terminal obtains voice information.

[0065] During the user's driving process, the in-vehicle terminal continuously monitors the surrounding environment's voice stream and acquires the voice information.

[0066] 304. When the voice information does not contain a wake-up word, the vehicle terminal recognizes the voice information and obtains the operation command.

[0067] When the voice information does not contain a wake-up word, after acquiring the voice information, the vehicle terminal converts the vocabulary in the voice information into computer-readable input, such as text content. After recognizing the voice information and obtaining the recognition result (i.e., the text content), the vehicle terminal searches for hot words matching the text content from a local hot word library to obtain the operation instruction matching the hot words.

[0068] 3041. The on-board terminal determines the execution end of the operation command.

[0069] It should be understood that the execution end of the operation command may be the in-vehicle terminal (e.g., the operation command is an instruction to control vehicle-related hardware devices) or the smart terminal (e.g., the operation command is an instruction to open an app installed on a mobile phone). Therefore, the in-vehicle terminal will determine the execution end of the operation command.

[0070] 3042. The vehicle-mounted terminal executes the operation instructions.

[0071] When the execution end of the operation command is the vehicle terminal, the vehicle terminal will directly execute the operation command.

[0072] 3043. The smart terminal executes the operation command.

[0073] When the execution end of the operation instruction is the smart terminal, the vehicle terminal forwards the operation instruction to the smart terminal, and the smart terminal executes the operation instruction after receiving it.

[0074] 305. When the voice message contains a wake word, the vehicle terminal sends the voice message to the cloud.

[0075] If the voice information contains a wake-up word, the vehicle terminal sends the voice information to the smart terminal, which then forwards it to the cloud. The cloud then recognizes the voice information and obtains the user's request.

[0076] It should be understood that in some embodiments, the voice information issued by the user begins with a specific wake word, such as the specific wake word "Hi, Siri" on Apple phones and the specific wake word "Xiaoyi Xiaoyi" on Huawei phones. After the vehicle terminal hears the wake word, the vehicle terminal begins to receive subsequent voice information from the user.

[0077] 306. The cloud recognizes voice information and obtains operation instructions.

[0078] The cloud platform converts the vocabulary in the voice information into computer-readable input, such as text content. After recognizing the voice information and obtaining the recognition result (i.e., the text content), the cloud platform searches for hot words that match the text content from a local hot word database to obtain the operation instructions corresponding to the hot words.

[0079] 3061. The intelligent terminal determines the execution end of the operation command.

[0080] Similarly, the execution end of the operation command may be the vehicle-mounted terminal or the smart terminal. Therefore, after the cloud sends the operation command to the smart terminal, the smart terminal will determine the execution end of the operation command.

[0081] 3062. The smart terminal executes the operation instructions.

[0082] When the execution end of the operation instruction is the smart terminal, the smart terminal directly executes the operation instruction after receiving it;

[0083] 3063. The vehicle-mounted terminal executes the operation command.

[0084] When the execution end of the operation instruction is the vehicle terminal, the smart terminal forwards the operation instruction to the vehicle terminal, and the vehicle terminal executes the operation instruction after receiving it.

[0085] It should be understood that, Figure 3In this process, steps 301 and 302 are hot word registration steps. This process is completed before voice information recognition and will not be repeated in subsequent voice recognition processes. However, the hot words in the vehicle terminal and the cloud can be updated. In some embodiments, the vehicle terminal and the cloud will periodically detect the usage frequency of the hot words they store within a certain period. If the usage frequency of a local hot word in the local hot word library within the target period is not greater than the target threshold, then the hot word can be considered a low-frequency hot word that the user has used less recently. In this case, the local hot word will be deleted from the local hot word library and stored in the cloud hot word library. If the usage frequency of a hot word in the cloud hot word library within the target period is greater than the target threshold, then the hot word can be considered a high-frequency hot word that the user has used more frequently recently. In this case, the hot word will be stored in the local hot word library.

[0086] In addition, it should be noted that during the specific identification process, Figure 3 Not all steps in the process need to be executed. Steps 303, 304, 3041, and 3042 can be considered as a complete voice information recognition and response process, corresponding to the application scenario where the user's voice information does not contain a wake word, and the execution end of the command corresponding to the voice information is the vehicle terminal. Similarly, steps 303, 304, 3041, and 3043 can also be considered as a complete voice information recognition and response process, corresponding to the application scenario where the user's voice information does not contain a wake word, and the execution end of the command corresponding to the voice information is the smart terminal. Furthermore, steps 303, 305, 306, 3061, and 3062 can be considered as a complete voice information recognition and response process. Their corresponding application scenario is that the user's voice information contains a wake-up word, and the execution end of the command corresponding to the voice information is the smart terminal. Steps 303, 305, 306, 3061, and 3063 can also be considered as a complete voice information recognition and response process. Their corresponding application scenario is that the user's voice information contains a wake-up word, and the execution end of the command corresponding to the voice information is the vehicle-mounted terminal. In other words, the specific execution terminal of the user's voice information can be divided into four cases based on the voice information recognition device and the operation command execution device, as detailed in Table 1 below:

[0087] Table 1

[0088]

[0089]

[0090] In one optional implementation, the vehicle-mounted terminal and the cloud are directly connected. When the voice information contains a wake-up word, after receiving the voice information, the vehicle-mounted terminal sends the voice information to the cloud. The cloud then converts the vocabulary in the voice information into computer-readable input, such as text content. After recognizing the voice information and obtaining the recognition result (i.e., the text content), the cloud searches for hot words matching the text content in its cloud hot word library to obtain the operation instruction, and sends the operation instruction to the vehicle-mounted terminal (without needing to determine the execution end of the operation instruction). The vehicle-mounted terminal then completes the corresponding operation according to the operation instruction. In some embodiments, if the storage and recognition performance of the vehicle-mounted terminal is higher than a certain level, the vehicle-mounted terminal can independently store hot words for voice information recognition. Therefore, if the voice information contains a wake-up word, the vehicle-mounted terminal can also obtain user needs and corresponding operation instructions by recognizing local hot words matching the voice information. The storage and recognition capabilities of the vehicle-mounted terminal are sufficient to independently complete the process of storing hot words and recognizing voice information. The system no longer needs to work in conjunction with the cloud, which reduces the data forwarding process and further improves the efficiency of voice information recognition.

[0091] In an optional implementation, when the storage and recognition performance of the vehicle-mounted terminal and the smart terminal are above a certain level, both the vehicle-mounted terminal and the smart terminal can independently complete the storage of hot words for voice information recognition. Therefore, in this embodiment, the storage and recognition performance of the vehicle-mounted terminal and the smart terminal are sufficient to independently complete the process of storing hot words for voice information recognition. At this time, the hot word library (hereinafter referred to as the smart instruction library) in the smart terminal also stores hot words related to operation instructions, and the system no longer needs to work in collaboration with the cloud. In this case, after acquiring the user's voice information, the vehicle-mounted terminal identifies local hot words matching the voice information to obtain the user's needs, obtains the operation instructions corresponding to the user's needs, and sends the operation instructions to the smart terminal. Alternatively, after acquiring the user's voice information, the vehicle-mounted terminal sends the voice information to the smart terminal. After receiving the voice information, the smart terminal identifies the voice information, obtains local hot words matching the voice information from the smart instruction library to obtain the user's needs, and obtains the operation instructions corresponding to the user's needs.

[0092] The following is a schematic diagram of the structure of a voice information recognition system provided in an embodiment of this application. Please refer to [link / reference]. Figure 4 The aforementioned speech recognition method can be based on Figure 4 The system implementation is shown below. Figure 4 As shown, the system includes:

[0093] The system includes an in-vehicle terminal 401, an intelligent terminal 402, and a cloud platform 403, wherein the intelligent terminal 402 is communicatively connected to the in-vehicle terminal 401 and the cloud platform 403.

[0094] The vehicle-mounted terminal 401 can be an in-vehicle device (such as an On-Board Unit (OBU)), a wireless terminal in self-driving vehicles, a terminal in transportation safety, etc. Any device capable of communicating with roadside equipment, vehicles, or vehicle management platforms falls within the protection scope of the vehicle-mounted terminal 401. The smart terminal 402 can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The cloud 403 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The smart terminal 402 and the cloud 403 can be directly or indirectly connected via wired or wireless communication; this application does not impose any restrictions on this connection. Specifically, the vehicle-mounted terminal 401, the smart terminal 402, and the cloud 403 can be the vehicle-mounted terminal, the smart terminal, and the cloud in the aforementioned voice information recognition method. In an optional implementation, the smart terminal 402 can communicate with the cloud 403 via a network. When executing the aforementioned voice information recognition method, the smart terminal 402 can transmit the aforementioned text information to the cloud 403 and receive operation instructions returned by the cloud 403. The network can be an external network, such as the Internet, or an internal network, such as Ethernet or a virtual private network. Optionally, the network can also be a wireless network, such as a WiFi network or a cellular network.

[0095] When performing the aforementioned data and voice information recognition method, the vehicle terminal 401 acquires the user's voice information and determines whether the voice information contains a wake-up word. If the voice information does not contain a wake-up word, the vehicle terminal 401 acquires the user's needs by recognizing local hot words that match the voice information and acquires operation instructions corresponding to the user's needs. If the voice information contains a wake-up word, the vehicle terminal 401 sends the voice information to the cloud 403. The cloud 403 receives the voice information, acquires the user's needs by recognizing cloud hot words that match the voice information, and sends operation instructions corresponding to the user's needs to the vehicle terminal 401. The vehicle terminal 401 executes the corresponding operation according to the acquired operation instructions.

[0096] Alternatively, the vehicle terminal 401 acquires the user's voice information and determines whether the voice information contains a wake-up word. If the voice information does not contain a wake-up word, the vehicle terminal 401 identifies local hot words matching the voice information to obtain the user's needs, acquires operation instructions corresponding to the user's needs, and sends the acquired operation instructions to the smart terminal 402. The smart terminal 402 executes the corresponding operation according to the operation instructions. If the voice information contains a wake-up word, the vehicle terminal 401 sends the voice information to the cloud 403. The cloud 403 receives the voice information, identifies cloud hot words matching the voice information to obtain the user's needs, and sends operation instructions corresponding to the user's needs to the smart terminal 402. The smart terminal 402 executes the corresponding operation according to the acquired operation instructions.

[0097] In an optional implementation, when the storage and recognition performance of the vehicle terminal 401 and the smart terminal 402 are above a certain level, both the vehicle terminal 401 and the smart terminal 402 can independently complete the storage of hot words for voice information recognition. Therefore, in this embodiment, the storage and recognition performance of the vehicle terminal 401 and the smart terminal 402 are sufficient to independently complete the process of storing hot words for voice information recognition. At this time, the hot word library (hereinafter referred to as the smart instruction library) in the smart terminal 402 also stores hot words related to operation instructions, and the system no longer needs to work collaboratively with the cloud. At this time, after the vehicle terminal 401 obtains the user's voice information, the vehicle terminal 401 obtains the user's needs by recognizing local hot words that match the voice information, obtains the operation instructions corresponding to the user's needs, and sends the operation instructions to the smart terminal 402. Alternatively, after acquiring the user's voice information, the vehicle terminal 401 sends the voice information to the smart terminal 402. After receiving the voice information, the smart terminal 402 recognizes the voice information, retrieves local hot words matching the voice information from the smart command library to obtain the user's needs, and obtains operation instructions corresponding to the user's needs.

[0098] In one optional implementation, the vehicle terminal 401 and the cloud 403 are directly communicatively connected. When the voice information contains a wake-up word, after receiving the voice information, the vehicle terminal 401 sends the voice information to the cloud 403, which converts the vocabulary in the voice information into computer-readable input, such as text content. After recognizing the voice information and obtaining the recognition result (i.e., the text content), the cloud 403 searches for hot words matching the text content in a cloud hot word database to obtain the operation instruction, and sends the operation instruction to the vehicle terminal 401. The vehicle terminal 401 completes the corresponding operation according to the operation instruction. In some embodiments, if the storage and recognition performance of the vehicle terminal 401 is higher than a certain level, the vehicle terminal 401 can independently complete the storage of hot words for voice information recognition. Therefore, if the voice information contains a wake-up word, the vehicle terminal 401 can also obtain user needs by recognizing local hot words that match the voice information and obtain operation instructions corresponding to the user needs. The system no longer needs to work in conjunction with the cloud 403, which reduces the data forwarding process and further improves the recognition efficiency of voice information.

[0099] In one optional implementation, the vehicle terminal 401 periodically detects the usage frequency of hot words in its local hot word library and the cloud hot word library in the cloud 403. If the usage frequency of a local hot word in the local hot word library within a target period is not greater than a target threshold, the local hot word is deleted from the local hot word library and stored in the cloud hot word library. The hot words in the cloud hot word library include those whose usage frequency is not greater than the target threshold. If the usage frequency of a cloud hot word in the cloud hot word library within a target period is greater than the target threshold, the cloud hot word is stored in the local hot word library. Any hot word in the local hot word library is one whose usage frequency is greater than the target threshold. This ensures that high-frequency hot words are responded to smoothly while reducing the possibility of false responses to low-frequency hot words, thus ensuring the continuous high efficiency of user voice information recognition.

[0100] In some embodiments, the hot words stored in the cloud-based hot word library include not only hot words whose frequency of use is below a target threshold (hereinafter referred to as low-frequency hot words), but also hot words whose frequency of use is above the target threshold (i.e., high-frequency hot words). That is, when a user cannot determine whether the hot word corresponding to their expected operation is a low-frequency hot word or a high-frequency hot word, they can directly speak voice information containing the wake word and the hot word, so that the voice information can be successfully responded to. Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 5 As shown, the electronic device 50 includes a processor 501, a memory 502, and a communication interface 503; the processor 501, the memory 502, and the communication interface 503 are interconnected via a bus 504. Figure 5 The electronic device in the embodiment can implement the voice information recognition method described above.

[0101] The memory 502 includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), or compact disc read-only memory (CDROM), and is used for related instructions and data. The communication interface 503 is used for receiving and sending data.

[0102] Processor 501 can be one or more central processing units (CPUs). If processor 501 is a CPU, the CPU can be a single-core CPU or a multi-core CPU. The steps performed by the voice information processing system in the above embodiment can be based on... Figure 5 The structure of the electronic device shown.

[0103] The processor 501 in the electronic device 50 is used to read the program code stored in the memory 502 and execute the voice information processing method in the foregoing embodiment.

[0104] In embodiments of this application, another computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, performs the following: acquiring a user's voice information; selecting different methods to acquire user needs based on whether the voice information contains a wake word; acquiring operation instructions based on the user needs; and performing corresponding operations based on the operation instructions.

[0105] This application provides a computer program product containing instructions that, when run on a computer, causes the computer to execute the voice information recognition method provided in the foregoing embodiments.

[0106] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method of recognizing voice information, characterized by, The method comprises the following steps: obtaining voice information of a user; in the case that the voice information does not contain a wake-up word, recognizing the voice information to obtain a user demand, and in the case that the voice information does not contain a wake-up word, recognizing a local hot word matched with the user demand to obtain an operation instruction corresponding to the local hot word, the local hot word being stored in a local hot word library; in the case that the voice information contains a wake-up word, receiving the user demand returned by a cloud, the user demand being obtained by recognizing the voice information by the cloud, the voice information being sent to the cloud by an intelligent terminal; in the case that the voice information contains a wake-up word, recognizing a cloud hot word matched with the user demand to obtain an operation instruction corresponding to the cloud hot word, the cloud hot word being stored in a cloud hot word library; performing a corresponding operation according to the operation instruction.

2. The method according to claim 1, wherein the performing a corresponding operation according to the operation instruction comprises: sending the operation instruction to an intelligent terminal; and the intelligent terminal performing a corresponding operation according to the operation instruction.

3. The method according to claim 1, further comprising: periodically detecting a usage frequency of a hot word in the local hot word library and the cloud hot word library; in the case that a usage frequency of a local hot word in the local hot word library in a target period is not greater than a target threshold, deleting the local hot word from the local hot word library and storing the hot word in the cloud hot word library, the hot words in the cloud hot word library including hot words with a usage frequency not greater than the target threshold; and in the case that a usage frequency of a cloud hot word in the cloud hot word library in a target period is greater than the target threshold, storing the cloud hot word in the local hot word library, any hot word in the local hot word library being a hot word with a usage frequency greater than the target threshold.

4. A speech information recognition system characterized by comprising: The system comprises a vehicle terminal, a cloud, wherein the vehicle terminal and the cloud are in communication connection, the vehicle terminal obtains voice information of a user and judges whether the voice information contains a wake-up word; if the voice information does not contain a wake-up word, the vehicle terminal obtains a user demand by recognizing a local hot word matched with the voice information and obtains an operation instruction corresponding to the user demand; if the voice information contains a wake-up word, the vehicle terminal sends the voice information to the cloud; the cloud receives the voice information, obtains a user demand by recognizing a cloud hot word matched with the voice information, and sends an operation instruction corresponding to the user demand to the vehicle terminal; the vehicle terminal performs a corresponding operation according to the obtained operation instruction.

5. The system according to claim 4, wherein if the voice information contains a wake-up word and the vehicle terminal supports independent completion of storage of a hot word and recognition of voice information, the vehicle terminal obtains a user demand by recognizing a local hot word matched with the voice information and obtains an operation instruction corresponding to the user demand. The system comprises a vehicle terminal, a cloud and an intelligent terminal, wherein the vehicle terminal, the cloud and the intelligent terminal are in communication connection, 6. A speech information recognition system characterized by comprising: ​ The vehicle terminal acquires voice information of a user; determines whether the voice information contains a wake-up word; If the voice information does not contain a wake-up word, the vehicle terminal acquires user demand by identifying a local hot word matched with the voice information, acquires an operation instruction corresponding to the user demand, and sends the operation instruction corresponding to the user demand to the intelligent terminal; If the voice information contains a wake-up word, the vehicle terminal sends the voice information to the intelligent terminal; the intelligent terminal sends the voice information to the cloud; the cloud receives the voice information, acquires user demand by identifying a cloud hot word matched with the voice information, and sends an operation instruction corresponding to the user demand to the intelligent terminal; The intelligent terminal executes corresponding operation according to the acquired operation instruction.

7. The system of claim 6, if the voice information contains a wake-up word, and the storage performance and identification performance of the vehicle terminal and the intelligent terminal both support independent completion of the process of storing hot words and identifying the voice information; The vehicle terminal acquires user demand by identifying a local hot word matched with the voice information, acquires an operation instruction corresponding to the user demand, and sends the operation instruction to the intelligent terminal; Or, The vehicle terminal sends the voice information to the intelligent terminal; the intelligent terminal acquires user demand by identifying a local hot word of the intelligent terminal matched with the voice information, and acquires an operation instruction corresponding to the user demand.

8. The system of claim 6, before the intelligent terminal executes corresponding operation according to the acquired operation instruction, The intelligent terminal determines an execution end of the operation instruction; The intelligent terminal executes corresponding operation according to the acquired operation instruction, including: In the case where the execution end is the intelligent terminal, the intelligent terminal executes corresponding operation according to the acquired operation instruction; In the case where the execution end is the vehicle terminal, the intelligent terminal sends the operation instruction to the vehicle terminal; the vehicle terminal executes corresponding operation according to the acquired operation instruction.

9. The system of claim 6, the local hot words are stored in a local hot word library of the vehicle terminal, any hot word in the local hot word library is a hot word with a user usage frequency greater than a target threshold; the cloud hot words are stored in a cloud hot word library of the cloud, the hot words in the cloud hot word library include instructions with a user usage frequency not greater than the target threshold; the vehicle terminal: Periodically detects the usage frequency of hot words in the local hot word library and the cloud hot word library; In the case where the usage frequency of a local hot word in the local hot word library in a target period is not greater than a target threshold, the local hot word is deleted from the local hot word library, and the hot word is stored in the cloud hot word library; In the case where the usage frequency of a cloud hot word in the cloud hot word library in a target period is greater than the target threshold, the cloud hot word is stored in the local hot word library.

10. An electronic device, comprising: Include: A memory for storing a program; a processor configured to execute the program stored by the memory, the processor being configured to execute the steps of the method of any of claims 1 to 3 when the program is executed.

11. A computer readable storage medium, characterized in that, a computer program stored in the computer readable storage medium, the computer program being configured to execute the steps of the method of any of claims 1 to 3 when the computer program is run on one or more processors.

Images