A speech recognition method, a speech recognition system, a speech recognition device, and a medium

Abstract

The application discloses a voice recognition method, a voice recognition system, a voice recognition device and a medium, and the method comprises the following steps: first, obtaining first audio data; the first audio data comprises at least one sound data; performing voiceprint recognition on the first audio data to obtain voiceprint data; wherein the voiceprint data corresponds to the sound data; shielding target sound data in the first audio data based on a preset voiceprint list according to the voiceprint data to obtain second audio data; performing voice conversion on the second audio data to obtain text data; performing target vocabulary matching based on a preset vocabulary list according to the text data; and responding to the second audio data when the target vocabulary does not exist in the text data. Through voiceprint recognition and voice conversion on the sound data in the audio data, and based on the preset voiceprint list and vocabulary list, the application realizes accurate recognition and response of the sound data, can effectively prevent mutual interference of voice interaction, and can be widely applied in the technical field of voice recognition.

Description

Technical Field

[0001] This invention relates to the field of speech recognition technology, and in particular to a speech recognition method, speech recognition system, speech recognition device, and medium. Background Technology

[0002] With the development of voice technology, more and more voice robots have emerged on the market. As voice robots become more widespread, the shortcomings of existing voice robot solutions have been exposed.

[0003] Existing voice robots inherently have a certain false wake-up rate. Inevitably, when multiple voice robots are awake, they will engage in continuous, disjointed conversations, negatively impacting the user experience. For example, in the same space, if there are two or more different voice assistants, and a user accidentally wakes up both, or if the first voice assistant is still listening when the second is woken up, the two voice assistants will be interfered with by each other's speech. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a speech recognition method, a speech recognition system, a speech recognition device, and a medium, which can accurately perform speech recognition.

[0005] On one hand, embodiments of the present invention provide a speech recognition method, including:

[0006] Acquire first audio data; wherein the first audio data includes at least one type of sound data;

[0007] Voiceprint recognition is performed on the first audio data to obtain voiceprint data; wherein, the voiceprint data corresponds to the sound data;

[0008] Based on the voiceprint data and a preset voiceprint list, the target sound data in the first audio data is filtered out to obtain the second audio data;

[0009] The second audio data is converted into speech to obtain text data;

[0010] Based on the text data and a pre-defined vocabulary list, target vocabulary matching is performed.

[0011] If the target word is not present in the text data, respond to the second audio data.

[0012] Optionally, it also includes:

[0013] Extract the sound data of the target object and train it to obtain a preset voiceprint list;

[0014] The preset voiceprint list includes a blacklist and a whitelist.

[0015] Optionally, it also includes:

[0016] Obtain the target vocabulary and organize it to obtain a preset vocabulary list;

[0017] The target vocabulary includes wake words.

[0018] Optionally, based on the voiceprint data and a preset voiceprint list, the target sound data in the first audio data is masked to obtain the second audio data, including:

[0019] Based on the voiceprint data, the target voiceprint data is determined by matching the voiceprints against a preset voiceprint list.

[0020] Based on the target voiceprint data, determine the target sound data in the first audio data that corresponds to the target voiceprint data, mask the target sound data in the first audio data, and use the remaining sound data as the second audio data.

[0021] In cases where the target voiceprint data is not present in the voiceprint data, the first audio data is used as the second audio data.

[0022] Optionally, voiceprint recognition is performed on the first audio data to obtain voiceprint data, including:

[0023] Voiceprint data is obtained by performing voiceprint recognition on the first audio data using voiceprint recognition technology.

[0024] The amount of sound data and voiceprint data is the same, and the voiceprint data is obtained based on the sound data.

[0025] Optionally, the second audio data is subjected to speech conversion, including:

[0026] The speech content of the second audio data is identified using speech recognition technology.

[0027] The text data is obtained by converting the speech content into text.

[0028] Optionally, it also includes:

[0029] If there is no sound data in the second audio data, or if the target word exists in the text data, clear the recognition data;

[0030] The identification data includes first audio data, second audio data, voiceprint data, and text data.

[0031] On the other hand, embodiments of the present invention provide a speech recognition system, including:

[0032] The first module is used to acquire first audio data; wherein the first audio data includes at least one type of sound data.

[0033] The second module is used to perform voiceprint recognition on the first audio data to obtain voiceprint data; wherein, the voiceprint data corresponds to the sound data.

[0034] The third module is used to filter out the target sound data in the first audio data based on the voiceprint data and a preset voiceprint list to obtain the second audio data.

[0035] The fourth module is used to convert the second audio data into speech data.

[0036] The fifth module is used to match target words based on the text data and a preset word list;

[0037] The sixth module is used to respond to the second audio data when the target word is not present in the text data.

[0038] On the other hand, embodiments of the present invention provide a voice recognition device, including a processor and a memory;

[0039] Memory is used to store programs;

[0040] The processor executes the program as described above.

[0041] On the other hand, embodiments of the present invention provide a computer-readable storage medium storing a program that is executed by a processor to implement the method described above.

[0042] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0043] This invention first acquires first audio data, which includes at least one type of sound data. Voiceprint recognition is then performed on the first audio data to obtain voiceprint data, which corresponds to the sound data. Based on the voiceprint data and a preset voiceprint list, target sound data in the first audio data is filtered out to obtain second audio data. The second audio data is then converted to text data. Based on the text data and a preset vocabulary list, target vocabulary matching is performed. If the target vocabulary is not present in the text data, a response is given to the second audio data. This invention, through voiceprint recognition and speech conversion of sound data in audio data, and based on preset voiceprint and vocabulary lists, achieves accurate recognition and response to sound data, effectively preventing interference between voice interactions. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a schematic diagram of the overall process of a speech recognition method provided in an embodiment of the present invention;

[0046] Figure 2 A schematic diagram illustrating the workflow of a speech recognition method provided in an embodiment of the present invention;

[0047] Figure 3 A schematic diagram of a speech recognition system provided in an embodiment of the present invention;

[0048] Figure 4 This is a schematic diagram of a speech recognition device provided in an embodiment of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0050] Current voice assistants have the following drawbacks: First, when two voice assistants are in the same space and both are in voice interaction mode, what they say to each other will be recorded by the other's voice robot, thus interfering with each other and affecting the recognition of the real user's voice; Second, existing voice assistants cannot exit the interaction when the user wakes up other voice assistants, thus indirectly causing the above problems.

[0051] In view of this, on the one hand, referring to Figure 1 The present invention provides a speech recognition method, comprising:

[0052] S100, Obtain the first audio data;

[0053] It should be noted that the first audio data includes at least one type of sound data.

[0054] Specifically, one type of sound data corresponds to the sound source of a target object, including the voice of the user or other voice assistants.

[0055] S200: Perform voiceprint recognition on the first audio data to obtain voiceprint data;

[0056] It should be noted that voiceprint data corresponds to sound data. In some embodiments, voiceprint recognition technology is used to perform voiceprint recognition on the first audio data to obtain voiceprint data; wherein, the amount of sound data and voiceprint data is the same, and the voiceprint data is obtained based on the sound data.

[0057] Specifically, voiceprint recognition is performed on each sound data point in the first audio data set, with a one-to-one correspondence between voiceprint data and sound data. Based on voiceprint recognition technology (identifying the speaker's identity, also known as speaker identification, a type of biometric technology), it can identify how many speakers there are (reference: the highest experimental accuracy rate for voiceprint recognition is currently 99.7%). The principle of voiceprint recognition is based on the characteristic that different people's voices have different formant distributions in the spectrogram (i.e., voiceprint data), comparing the voices of two people and determining whether they are the same person by comparing the pronunciation of the same phonemes.

[0058] S300: Based on the voiceprint data and a preset voiceprint list, the target sound data in the first audio data is masked to obtain the second audio data;

[0059] It should be noted that in some embodiments, voiceprint matching is performed based on a preset voiceprint list according to the voiceprint data to determine the target voiceprint data; according to the target voiceprint data, the target sound data corresponding to the target voiceprint data in the first audio data is determined, the target sound data in the first audio data is masked, and the remaining sound data is used as the second audio data; wherein, when there is no target voiceprint data in the voiceprint data, the first audio data is used as the second audio data.

[0060] In some embodiments, the method further includes: extracting the sound data of the target object and training a preset voiceprint list; wherein the preset voiceprint list includes a blacklist list and a whitelist list.

[0061] Specifically, high accuracy is achieved by pre-recording the voices of the targets to be blocked and training them. For example, voices of voice assistants from other products are included before shipment (resulting in a blacklist of preset voiceprints) to prevent interference from voice assistant interactions. When a target voiceprint data matching the preset voiceprint list (blacklist) appears in the voiceprint data, its corresponding voice data is blocked, thus obtaining the second audio data.

[0062] When there is no remaining sound data in the first audio data after the target voiceprint data is blocked, the response / interaction will exit directly, and the previously identified first audio data, voiceprint data, etc. will be cleared.

[0063] Furthermore, to avoid situations where a user's voice is very similar to that of a voice assistant, potentially leading to misidentification, users are provided with optional actions. For example, before shipment, a batch of existing voice robots on the market are identified, marked, and displayed, allowing users to choose which voice robots they wish to block, enabling the creation of a custom preset voiceprint list. Simultaneously, by recording the user's voiceprint data into the system and marking it (including it in a whitelist), the probability of being misidentified as a voice assistant can be reduced, improving recognition accuracy.

[0064] S400: Perform speech conversion on the second audio data to obtain text data;

[0065] It should be noted that in some embodiments, the speech content of the second audio data is identified using speech recognition technology; text data is obtained by converting the speech content into text.

[0066] Specifically, based on speech recognition technology (which identifies the content of speech and automatically converts human speech into corresponding text or executes related commands using a computer, such as voice input for text or smart speakers; for reference, the highest word accuracy in current experimental data is 97%), the second audio data is converted into text data through speech-to-text conversion.

[0067] S500: Based on the text data and a preset vocabulary list, perform target word matching;

[0068] In some embodiments, the method further includes: acquiring target words and organizing the target words to obtain a preset word list; wherein the target words include wake words.

[0069] Specifically, high accuracy is achieved by pre-recording wake words from other voice assistants and training them. For example, wake words from certain product voice assistants may not be included before shipment (resulting in a preset vocabulary list). Combining speech recognition technology with the preset vocabulary list, it is possible to identify whether specific words (such as wake words from other voice assistants) exist in the text data.

[0070] Specifically, when a target word matching a preset vocabulary list appears in the text data, it is interpreted as an interaction with another voice assistant, thus triggering the user's own interaction and clearing the previously acquired first audio data, second audio data, voiceprint data, and text data. Simultaneously, users can customize and input other wake words, which are automatically trained by the cloud server to create a customized preset vocabulary list.

[0071] S600. When the target word is not present in the text data, respond to the second audio data;

[0072] Specifically, the acquired first audio data is then processed through voiceprint recognition and speech conversion. Based on the voiceprint data matching, the voice data of other voice assistants is masked, and target word matching is used to prevent interaction interference caused by false wake-ups. This allows for accurate interactive responses to the second audio data, which contains no interfering elements.

[0073] The technical principles of the present invention will be further explained below with reference to specific embodiments. The following is an explanation of the present invention and should not be regarded as a limitation thereof:

[0074] First, we need to do some preliminary preparations:

[0075] 1) Preliminary preparation: Masking the voice of the voice robot (i.e., pre-setting a voiceprint list):

[0076] ① Record the voices of other voice assistants in advance and train them to achieve high accuracy. For example, include the voice assistant voices of certain products before they leave the factory.

[0077] ② Users can input custom voices from other voice robots not already in the existing system, and the cloud server will automatically train them.

[0078] ③ To avoid situations where a user's voice is very similar to that of a voice assistant, potentially leading to misidentification, users are provided with optional actions. For example, before shipment, the voiceprints of a batch of existing voice robots on the market are identified, marked, and displayed, allowing users to choose which voice robots they wish to block.

[0079] ④ By recording their own / family / friends' voices into the system and marking them (including them in the whitelist), users can greatly reduce the probability of being mistaken for a voice assistant.

[0080] 2) Preliminary preparation: blocking the wake-up words of the voice robot (preset vocabulary list):

[0081] ① Record wake words from other voice assistants in advance and train them to achieve high accuracy. For example, wake words for some product voice assistants may not be included before the product leaves the factory.

[0082] ② Users can input other wake words by customization, and the cloud server will automatically perform training.

[0083] like Figure 2 As shown, the workflow is as follows:

[0084] When a user inputs voice, the voiceprint recognition module determines whether the speaker is a "specific person". If so, it automatically masks the speaker's speech and only processes the voice of non-"specific persons".

[0085] If the voice only contains the voice of a "specific person", a specific reply will be given (either replying or not replying can be customized), and the voice interaction state will be automatically exited. If the voice robot containing only the voice of "B" is detected, the automatic reply will be: "B, it's such a coincidence to meet you here. See you later, bye-bye" (either replying or not replying can be customized).

[0086] When the speech recognition module detects a "specific word," it can first respond with a specific message (either a response or no response can be customized), and then automatically exit the voice interaction interface. For example, if the speech robot recognizes the user saying "Hello, B," it will reply, "I'll take my leave now," and then exit the voice interaction.

[0087] The system responds to the input speech when there is no audio of a "specific person" or when there is other audio besides the audio of a "specific person" and the speech recognition fails to recognize the "specific word".

[0088] In summary, this invention provides a speech recognition method that avoids interference between two or more voice robots, significantly improving the user experience. It also provides a way to automatically exit the voice interaction state when a user wakes up another voice assistant. Furthermore, it allows users to customize the "specific person" and "specific word" to be blocked, making it more intelligent. This invention solves the problem of two voice robots interfering with each other's dialogue; it also addresses the risk of increased interference when another voice robot is being woken up while the current robot is still in a voice interaction state (specifically, listening, parsing, or broadcasting). It achieves accurate recognition and response to sound data, effectively preventing mutual interference in voice interactions.

[0089] On the other hand, embodiments of the present invention provide a speech recognition system 700, comprising: a first module 710 for acquiring first audio data, wherein the first audio data includes at least one type of sound data; a second module 720 for performing voiceprint recognition on the first audio data to obtain voiceprint data, wherein the voiceprint data corresponds to the sound data; a third module 730 for filtering out target sound data in the first audio data based on the voiceprint data and a preset voiceprint list to obtain second audio data; a fourth module 740 for performing speech conversion on the second audio data to obtain text data; a fifth module 750 for performing target word matching based on the text data and a preset word list; and a sixth module 760 for responding to the second audio data when the target word is not present in the text data.

[0090] The content of the method embodiments of the present invention is applicable to the system embodiments. The specific functions implemented in the system embodiments are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0091] Another aspect of the present invention provides a voice recognition device 800, including a processor 810 and a memory 820;

[0092] The 820 memory is used to store programs;

[0093] The 810 processor executes programs as described above.

[0094] The content of the method embodiments of the present invention is applicable to the device embodiments. The specific functions implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0095] Another aspect of this invention provides a computer-readable storage medium storing a program that is executed by a processor to implement the method described above.

[0096] The content of the method embodiments of the present invention is applicable to the computer-readable storage medium embodiments. The specific functions implemented by the computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0097] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0098] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is altered and sub-operations described as part of a larger operation are executed independently.

[0099] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the described functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0100] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0101] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution means, apparatus, or device (such as a computer-based device, a processor-including device, or other means that can fetch and execute instructions from, or in conjunction with, an instruction execution means, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution means, apparatus, or device.

[0102] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0103] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution device. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0104] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0105] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0106] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of the present invention.

Claims

1. A speech recognition method, characterized in that, include: Acquire first audio data; wherein the first audio data includes at least one type of sound data; Voiceprint recognition is performed on the first audio data to obtain voiceprint data; wherein the voiceprint data corresponds to the sound data; Based on the voiceprint data and a preset voiceprint list, target sound data in the first audio data is filtered out to obtain the second audio data; The second audio data is converted into text data through speech conversion. Based on the text data, target words are matched against a preset vocabulary list; If the target word is not present in the text data, respond to the second audio data; The step of obtaining second audio data by filtering out target sound data in the first audio data based on the voiceprint data and a preset voiceprint list includes: Based on the voiceprint data, voiceprint matching is performed on a preset voiceprint list to determine the target voiceprint data; Based on the target voiceprint data, determine the target sound data in the first audio data that corresponds to the target voiceprint data, block the target sound data in the first audio data, and use the remaining sound data as the second audio data. Wherein, if the target voiceprint data is not present in the voiceprint data, the first audio data is used as the second audio data; Specifically, if there is no remaining sound data in the first audio data after the target voiceprint data is blocked, the response will exit directly, and the previously identified first audio data and voiceprint data will be cleared. When the text data contains a target word that matches the preset word list, it is determined to be an interaction action with other voice assistants, and the interaction will exit, clearing the previously identified first audio data, second audio data, voiceprint data and text data.

2. The speech recognition method according to claim 1, characterized in that, Also includes: Extract the sound data of the target object and train it to obtain a preset voiceprint list; The preset voiceprint list includes a blacklist and a whitelist.

3. The speech recognition method according to claim 1, characterized in that, Also includes: Obtain the target vocabulary and organize it into a preset vocabulary list; The target vocabulary includes wake words.

4. The speech recognition method according to claim 1, characterized in that, The step of performing voiceprint recognition on the first audio data to obtain voiceprint data includes: Voiceprint data is obtained by performing voiceprint recognition on the first audio data using voiceprint recognition technology. The amount of sound data and voiceprint data is the same, and the voiceprint data is obtained based on the sound data.

5. The speech recognition method according to claim 1, characterized in that, The process of converting the second audio data into speech includes: The speech content of the second audio data is identified using speech recognition technology; The speech content is converted into text to obtain text data.

6. A speech recognition system, characterized in that, The system, applied to the method of claim 1, comprises: The first module is used to acquire first audio data; wherein the first audio data includes at least one type of sound data. The second module is used to perform voiceprint recognition on the first audio data to obtain voiceprint data; wherein the voiceprint data corresponds to the sound data; The third module is used to filter out target sound data in the first audio data based on the voiceprint data and a preset voiceprint list to obtain the second audio data. The fourth module is used to convert the second audio data into speech data to obtain text data; The fifth module is used to perform target word matching based on the text data and a preset word list; The sixth module is used to respond to the second audio data when the target word is not present in the text data.

7. A voice recognition device, characterized in that, Including the processor and memory; The memory is used to store programs; The processor executes the program to implement the method as described in any one of claims 1 to 5.

8. A computer storage medium, characterized in that, The storage medium stores a program that is executed by a processor to implement the method as described in any one of claims 1 to 5.

Images