Speech recognition method and device of equipment, equipment and storage medium

Abstract

The embodiment of the invention provides a voice recognition method and device for equipment, the equipment and a storage medium, and the method comprises the steps: carrying out the voiceprint recognition in response to a voice wake-up instruction, and obtaining the voiceprint information of a user; according to a matching result of the user voiceprint information and preset voiceprint information, obtaining target language type coding information of the preset voiceprint information; the target language type coding information is transmitted to a cloud end, and the cloud end switches a language recognition model of the equipment into a target language recognition model of the target language type based on the target language type coding information; and performing voice broadcasting according to the target language type. The voice recognition technology and the voiceprint recognition technology are adopted to analyze and recognize the voice audio of the user, so that the voiceprint information of the user is recognized, the target language type correspondingly needed to be used by the equipment is obtained, then the target language type is automatically selected and used for voice interaction, and the switching accuracy under the condition of multiple languages is improved.

Description

Technical Field

[0001] The present invention relates to the field of device control technology, and in particular to a device speech recognition method, a device speech recognition device, a corresponding electronic device, and a corresponding computer-readable storage medium. Background Art

[0002] With the development of AI technology, voice technology is gradually being widely used in the home field. More and more home products can be operated through voice interaction to meet users' needs for fast services. With the application of voice technology in home scenes, the demand for supported languages ​​has increased, especially when a family may have multiple dialects.

[0003] However, once the user needs to switch the language, he or she usually needs to enter the voice recognition engine setting interface through the APP or mini program and manually switch to the dialect to realize the voice recognition of home appliances. The aforementioned voice switching method has low switching efficiency and is only applicable to products equipped with APP and mini programs, which is not smart enough. Summary of the invention

[0004] In view of the above problems, embodiments of the present invention are proposed to provide a speech recognition method for a device, a speech recognition apparatus for a device, a corresponding electronic device, and a corresponding computer-readable storage medium that overcome the above problems or at least partially solve the above problems.

[0005] The embodiment of the present invention discloses a speech recognition method for a device, the method comprising:

[0006] Respond to the voice wake-up command and perform voiceprint recognition to obtain the user's voiceprint information;

[0007] According to the matching result between the user voiceprint information and the preset voiceprint information, obtaining the target language type encoding information of the preset voiceprint information;

[0008] Transmitting the target language encoding information to the cloud, and the cloud switching the language recognition model of the device to the target language recognition model of the target language based on the target language encoding information;

[0009] The voice broadcast is performed according to the target language.

[0010] Optionally, the preset voiceprint information is stored in a preset storage repository, in which a binding relationship between a voiceprint information code of the preset voiceprint information and language type code information of a preset language type is pre-stored;

[0011] The step of obtaining target language encoding information of the preset voiceprint information according to a matching result between the user voiceprint information and the preset voiceprint information includes:

[0012] In response to the user voiceprint information matching the preset voiceprint information, based on the binding relationship, the target language type encoding information of the preset voiceprint information is obtained from the preset storage repository.

[0013] Optionally, the step of obtaining target language encoding information of the preset voiceprint information in response to a matching result between the user voiceprint information and the preset voiceprint information further includes:

[0014] In response to the absence of preset voiceprint information matching the user voiceprint information in the preset storage repository, the user voiceprint information is stored in the preset storage repository.

[0015] Optionally, storing the user voiceprint information in the preset storage repository includes:

[0016] Generate a voiceprint information code for the user's voiceprint information and prompt language selection information;

[0017] In response to the user's language selection instruction, language type coding information for the selected language type is generated, and the language type coding information of the selected language type and the voiceprint information coding of the user's voiceprint information are stored in the preset storage repository.

[0018] Optionally, the responding to the voice wake-up instruction and performing voiceprint recognition to obtain user voiceprint information includes:

[0019] Detecting the presence of a preset wake-up word in the user's voice audio, switching the voice recognition module of the device from a sleep mode to a wake-up state;

[0020] The voiceprint recognition is performed on the user's voice audio through the voice recognition module in the awake state to obtain the user's voiceprint information.

[0021] Optionally, when the speech recognition module is in an awake state, the method further includes:

[0022] In response to a user's language switching instruction, obtain the language type to be switched, and obtain language type selection information of the language type to be switched;

[0023] When there is preset voiceprint information matching the user's voiceprint information in the preset storage repository, the voiceprint information code of the preset voiceprint information is obtained, and the language type selection information of the language to be switched is bound to the voiceprint information code of the preset voiceprint information and stored in the preset storage repository.

[0024] Optionally, the method further includes:

[0025] When there is no preset voiceprint information matching the user's voiceprint information in the preset storage repository, generating a voiceprint information code for the user's voiceprint information, and prompting language selection information;

[0026] In response to the user's language selection instruction, language type coding information for the selected language type is generated, and the language type coding information of the selected language type and the voiceprint information coding of the user's voiceprint information are stored in the preset storage repository.

[0027] Optionally, the step of storing the language type encoding information of the selected language type and the voiceprint information encoding of the user voiceprint information in the preset storage repository includes:

[0028] The voiceprint information encoding of the user's voiceprint information and the language type encoding information of the selected language type are bound and stored in the preset storage repository.

[0029] Optionally, the method further includes:

[0030] If no language selection instruction from the user is received within a preset time period, the preset language type is determined to be the language type selected for the user's voiceprint information.

[0031] The embodiment of the present invention further discloses a speech recognition device of a device, the device comprising:

[0032] The voiceprint recognition module is used to respond to the voice wake-up command and perform voiceprint recognition to obtain the user's voiceprint information;

[0033] A voiceprint matching module, configured to obtain target language encoding information of the preset voiceprint information according to a matching result between the user voiceprint information and the preset voiceprint information;

[0034] A language recognition model switching module, used for transmitting the target language type encoding information to the cloud, and the cloud switches the language recognition model of the device to a target language recognition model of the target language type based on the target language type encoding information;

[0035] The voice playing module is used for performing voice broadcasting in the target language.

[0036] Optionally, the preset voiceprint information is stored in a preset storage repository, in which a binding relationship between the voiceprint information encoding of the preset voiceprint information and the language type encoding information of a preset language type is pre-stored; the voiceprint matching module includes:

[0037] The target language encoding information acquisition submodule is used to respond to the user voiceprint information matching the preset voiceprint information and acquire the target language encoding information of the preset voiceprint information from the preset storage repository based on the binding relationship.

[0038] Optionally, the voiceprint matching module further includes:

[0039] The voiceprint information storage submodule is used for storing the user voiceprint information in the preset storage repository in response to the absence of preset voiceprint information matching the user voiceprint information in the preset storage repository.

[0040] Optionally, the voiceprint information storage submodule includes:

[0041] The voiceprint information storage unit is used to generate a voiceprint information code for the user's voiceprint information and prompt language type selection information; in response to the user's language type selection instruction, generate language type coding information for the selected language type, and store the language type coding information of the selected language type and the voiceprint information code of the user's voiceprint information in the preset storage repository.

[0042] Optionally, the voiceprint information storage unit includes:

[0043] The voiceprint information storage subunit is used to bind the voiceprint information code of the user's voiceprint information and the language type code information of the selected language type and store them in the preset storage library.

[0044] Optionally, the voiceprint information storage submodule includes:

[0045] The language type setting unit is used to determine the preset language type as the language type selected for the user voiceprint information if no language type selection instruction from the user is received within a preset time period.

[0046] Optionally, the voiceprint recognition module includes:

[0047] The voiceprint recognition submodule is used to detect the presence of a preset wake-up word in the user's voice audio, and switch the voice recognition module of the device from a sleep mode to a wake-up state; through the voice recognition module in the wake-up state, voiceprint recognition is performed on the user's voice audio to obtain the user's voiceprint information.

[0048] Optionally, when the speech recognition module is in the awake state, the device further includes:

[0049] A language switching module is used to respond to the user's language switching instruction, obtain the language type to be switched, and obtain the language type selection information of the language type to be switched; when there is preset voiceprint information matching the user's voiceprint information in the preset storage repository, obtain the voiceprint information code of the preset voiceprint information, bind the language type selection information of the language type to be switched with the voiceprint information code of the preset voiceprint information and store them in the preset storage repository.

[0050] Optionally, the language switching module is also used to generate a voiceprint information code for the user's voiceprint information and prompt language type selection information when there is no preset voiceprint information matching the user's voiceprint information in the preset storage repository; respond to the user's language type selection instruction, generate language type coding information for the selected language type, and store the language type coding information of the selected language type and the voiceprint information code of the user's voiceprint information in the preset storage repository.

[0051] An embodiment of the present invention further discloses an electronic device, comprising: a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein when the computer program is executed by the processor, any speech recognition method of the device is implemented.

[0052] The embodiment of the present invention further discloses a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the speech recognition method of any one of the devices described above is implemented.

[0053] The embodiments of the present invention include the following advantages:

[0054] In the embodiment of the present invention, the user's voiceprint information is obtained by performing voiceprint recognition in response to a voice wake-up command, and the target language type encoding information of the preset voiceprint information is obtained according to the matching result of the user's voiceprint information and the preset voiceprint information, and the obtained target language type encoding information is transmitted to the cloud, so that the cloud can switch the language recognition model of the device to the language recognition model of the target language type based on the target language type encoding information, so that the device switched to the target language recognition model can perform voice broadcasting in the target language type. By adopting voice recognition technology and voiceprint recognition technology, the user's voice audio is analyzed and recognized to recognize the user's voiceprint information, and the target language type that the device needs to use is obtained, and then the target language type is automatically selected and used for voice interaction, which improves the switching accuracy in multiple language situations, thereby realizing flexible switching of multiple language types, meeting the interaction needs of different groups of people for language types, and can be applicable to voice interaction products in multiple language types. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1is a schematic diagram of a framework of a device speech recognition system provided by an embodiment of the present invention;

[0056] Figure 2 A flowchart of a speech recognition method embodiment of a device of the present invention;

[0057] Figure 3 is a flowchart of another embodiment of a speech recognition method of a device of the present invention;

[0058] Figure 4 is a schematic diagram of a process of speech recognition performed by a device provided in an embodiment of the present invention;

[0059] Figure 5 It is a structural block diagram of an embodiment of a speech recognition device of a device of the present invention. DETAILED DESCRIPTION

[0060] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0061] With the application of voice technology in home scenes, the demand for supported language types has become higher. In the related technology of voice interaction of devices, as an example, it is manifested as identifying the characteristics of at least two language categories based on voice information, determining the matching degree with the language category based on the degree of similarity, and realizing the recognition of the target language category; as another example, it is manifested as identifying the language type that the terminal needs to use by the user's voice information, automatically selecting and using the language type for broadcasting, so as to realize automatic switching of different language types according to the language types used by different users, without the need for the user to manually select and set the language type for broadcasting; as another example, it is manifested as using the language detection module of the terminal device and the main wake-up word module in the voice wake-up module of the terminal device to detect and wake up the input voice information respectively, and matching the language type through the pronunciation characteristics. However, in the aforementioned voice recognition method, the recognition of the language type is either determined only by voice information or matched and determined by pronunciation characteristics. There are errors in the matching of language types, and when the number of language types increases and / or the characteristics of the language types are similar, matching errors are prone to occur.

[0062] The embodiment of the present invention adopts speech recognition technology and voiceprint recognition technology to analyze and identify user voice audio to identify user voiceprint information, obtain the target language type that the device needs to use, and then automatically select and use the target language type for voice interaction, thereby improving the switching accuracy in multiple language situations, thereby achieving flexible switching of multiple language types, meeting the interaction needs of different groups of people for language types, and being applicable to voice interaction products in multiple language types; further, by adding the application of voiceprint technology, based on the uniqueness of user voiceprint information and language type encoding information that is bound to user voiceprint information, using user voiceprint information as the basis for identifying the language type, it is possible to improve the switching accuracy in multiple language situations, thereby improving the user voice interaction experience.

[0063] Reference Figure 1 , shows a schematic diagram of the framework of a device voice recognition system provided by an embodiment of the present invention, wherein the device voice recognition system 110 involves a device end 11 and a cloud end 12.

[0064] Among them, the device end 11 may include a voice acquisition module, a voice recognition module, a voice broadcast module, and an information storage module, etc. The voice acquisition module is mainly responsible for capturing the sound signal in the environment and converting it into an electrical signal. For example, the capture of the sound signal and the conversion of the electrical signal can be realized by a microphone array; the voice recognition module can be mainly used to recognize the voice, including but not limited to the recognition of the language type and the user's voiceprint; the voice broadcast module mainly converts text information or other information into a voice signal for providing feedback or information; the information storage module can be mainly used to store the information obtained from the voice recognition module, and other related data, such as the user's voiceprint information and the language type encoding information in the embodiment of the present invention, etc., which are not limited in the embodiment of the present invention. In practical applications, the realization of the device's recognition of voice requires that the voice recognition module is in an online networking state, that is, the voice recognition module in an offline state can be awakened locally, so that the voice recognition module in the awakened state can call multiple modules such as the voice acquisition module, the voice recognition module, the voice broadcast module, and the information storage module to realize the device's recognition of the language type of the voice.

[0065] The cloud 112 may include a variety of language recognition models, which may be used to perform language recognition on different language types, including recognition of different dialects. Taking dialect recognition as an example, the cloud may include as many dialect recognition models as possible for recognizing different types of dialects, so as to cover the recognition of all dialects as much as possible, thereby achieving flexible switching of multiple languages, meeting the language interaction needs of different groups of people, and being applicable to voice interaction products in multiple languages.

[0066] In actual applications, the device voice recognition system 110 can be deployed on voice-controlled home devices. Taking the home device as a voice air-conditioning device as an example, the deployment of the device voice recognition system 110 on the voice air-conditioning device can be manifested as adding the above-mentioned voice acquisition module, voice recognition module, voice broadcast module and information storage module to the voice air-conditioning device to realize voice control of the air-conditioning. Specifically, it can be manifested as after the air-conditioning is connected to the Internet, the user can interact with the air-conditioning through dialogue, music playback, story playback, dialogue query, etc., so that the air-conditioning after the device voice recognition system is deployed has the same functions as the smart speaker, giving the user the same user experience as the smart speaker. In addition, since the device voice recognition system deployed by the voice air-conditioning device includes multiple language recognition models, the voice air-conditioning device can flexibly switch between multiple languages, thereby meeting the interaction needs of different groups of people for language types.

[0067] Reference Figure 2 , shows a flowchart of a speech recognition method embodiment of a device of the present invention, which is applied to a device end and may specifically include the following steps:

[0068] Step 201, responding to a voice wake-up command, performing voiceprint recognition to obtain user voiceprint information;

[0069] The device can be in sleep mode when not in use. If a user's voice wake-up command is detected, the device can enter the wake-up state from sleep mode. The device in the wake-up state can work normally and realize voice recognition.

[0070] In an embodiment of the present invention, the speech recognition performed by the device may be specifically manifested as recognition of the language type.

[0071] Among them, the voice wake-up instruction can be generated based on the detected user voice audio with a preset wake-up word. In one embodiment of the present invention, in order to identify the language type of the user who currently wakes up the device, the voiceprint recognition of the user who currently wakes up the device can be performed to obtain unique information for identifying the user and identify the language type of the user.

[0072] Step 202, obtaining target language encoding information of the preset voiceprint information according to the matching result between the user voiceprint information and the preset voiceprint information;

[0073] Specifically, for the user voiceprint information of a certain user, there may be a language type pre-set for the user's voiceprint information. At this time, after performing voiceprint recognition on the user who currently wakes up the device and obtaining the user's voiceprint information, the user's voiceprint information can be matched with the preset voiceprint information to obtain a matching result, and subsequent language type identification is performed based on the matching result.

[0074] The preset voiceprint information may be stored in a preset storage repository, and the preset voiceprint information in the preset storage repository has a preset language type by default.

[0075] Different voiceprint information is unique information used to identify different users, and voiceprint information can be used as a basis for switching languages.

[0076] In one case, if there is a preset voiceprint information matching the user's voiceprint information in the preset repository, it means that the user who currently wakes up the device has a preset language type, so that the device can recognize and respond to the instructions of the user who currently wakes up the device according to the set language type; in another case, if there is no preset voiceprint information matching the user's voiceprint information in the preset repository, it means that the user who currently wakes up the device does not have a preset language type. At this time, the language type can be set for the user voiceprint information of the user who currently wakes up the device and stored in the preset repository, so that when the user subsequently interacts with the device through voice, the language type set by the user can be obtained directly based on the user voiceprint information and the preset voiceprint information.

[0077] The information storage in the preset repository can be implemented based on the information storage module on the device side, which can be local storage. The language type set for a user's voiceprint information stored in the preset repository can be implemented based on the target language type encoding information of the voiceprint information.

[0078] Step 203, transmitting the target language encoding information to the cloud;

[0079] In actual applications, the device's recognition of user commands can be manifested as identifying the language type of the corresponding voice commands through a language recognition model, and then identifying specific voice information. The language recognition model is deployed in the cloud. Since the voice recognition on the device side depends on the language recognition model, the device side can transmit the target language type encoding information to the cloud at this time, so that the cloud side can switch the device's language recognition model to the target language type based on the target language type encoding information, thereby realizing flexible switching of multiple languages.

[0080] Switching the language recognition model of the device to the target language recognition model specifically means that the language recognition model currently used by the device is the target language recognition model, so as to switch the control device to the target language type, so that the device can perform voice recognition through the switched language recognition model, and realize voice interaction with the user using the language type.

[0081] Step 204: Perform voice broadcast according to the target language.

[0082] The device side whose language recognition model is switched to the target language recognition model can use this language type for voice interaction with the user. At this time, after receiving the user's voice command, the device side can use the target language recognition model to recognize the voice command, and after responding to the aforementioned voice command to execute the corresponding operation / control, it can broadcast voice to the user in the target language type, thereby realizing voice interaction with the user using this language type.

[0083] Exemplarily, the voice broadcast performed by the user may be a voice broadcast of the completion of a command, or an audio broadcast of a conversation with the user, etc., which is not limited in this embodiment of the present invention.

[0084] In one embodiment of the present invention, step 201, responding to a voice wake-up instruction, performing voiceprint recognition to obtain user voiceprint information, includes:

[0085] When a preset wake-up word is detected in the user's voice audio, the voice recognition module of the device is switched from the sleep mode to the wake-up state; the voice recognition module in the wake-up state performs voiceprint recognition on the user's voice audio to obtain the user's voiceprint information.

[0086] The device can be in sleep mode when not in use. Specifically, the voice recognition module of the device is in sleep mode. When the voice recognition module in sleep mode detects a preset wake-up word in the user's voice audio, it can wake up the voice recognition module, and at the same time, it can wake up the voice collection module, voice recognition module, voice broadcast module and information storage module, etc.; if the voice recognition module in sleep mode does not detect the preset wake-up word, it can maintain sleep mode and continue waiting.

[0087] Optionally, since the voice recognition module in sleep mode needs to rely on the user voice audio collected by the voice collection module to detect the preset wake-up word, when the voice recognition module is in sleep mode, the voice collection module is also in sleep mode. The voice collection module in sleep mode can collect user voice audio in a low power consumption state. The difference between the low power consumption state and the normal working state can be manifested in that the voice collection module in the low power consumption state can collect user voice audio within a preset range and / or user voice audio with a volume greater than a preset volume threshold. The voice collection conditions of the voice collection module in the normal working state may not be limited to the range and volume, and the embodiments of the present invention are not limited to this.

[0088] After being awakened, the voice recognition module can perform voiceprint recognition on the user's voice audio to obtain the user's voiceprint information.

[0089] In an embodiment of the present invention, step 202, obtaining target language encoding information of the preset voiceprint information according to the matching result between the user voiceprint information and the preset voiceprint information, includes:

[0090] In response to the user voiceprint information matching the preset voiceprint information, based on the binding relationship, the target language type encoding information of the preset voiceprint information is obtained from the preset storage repository.

[0091] Different voiceprint information is unique information used to identify different users, and voiceprint information can be used as a basis for switching languages.

[0092] In one case, if there is a preset voiceprint information matching the user's voiceprint information in the preset storage repository, it means that the user who currently wakes up the device has a preset language type, so that the device side can recognize and respond to the instructions of the user who currently wakes up the device according to the set language type.

[0093] Specifically, the preset voiceprint information is stored in a preset storage repository, and the language type set for a certain user's voiceprint information stored in the preset storage repository can be implemented based on the target language type coding information of the voiceprint information. That is, the voiceprint information coding of the preset voiceprint information pre-stored in the preset storage repository can have a binding relationship with the language type coding information of the preset language type. At this time, if the user's voiceprint information matches the preset voiceprint information, the language type coding information corresponding to this voiceprint information can be searched in the storage area, which can be specifically manifested as obtaining the target language type coding information of the preset voiceprint information from the preset storage repository based on the binding relationship, and then obtaining the language type of the user who currently wakes up the device.

[0094] In another case, in response to the absence of preset voiceprint information matching the user voiceprint information in the preset repository, the user voiceprint information is stored in the preset repository. That is, if there is no preset voiceprint information matching the user voiceprint information in the preset repository, it means that the user who currently wakes up the device does not have a preset language type. At this time, the language type can be set for the user voiceprint information of the user who currently wakes up the device and stored in the preset repository, so that when the user subsequently interacts with the device end by voice, the language type set by the user can be directly obtained based on the user voiceprint information and the preset voiceprint information.

[0095] Specifically, the information storage in the preset repository can be implemented based on the information storage module of the device end, and the information storage module can be local storage. The language type set for a certain user's voiceprint information stored in the preset repository can be implemented based on the target language type encoding information of the voiceprint information. Specifically, it can be manifested as generating a voiceprint information encoding for the user's voiceprint information, and then establishing a binding relationship between the voiceprint information encoding of the user's voiceprint information and the language type encoding information of the language type to be set, and storing it in the preset repository, so that when the user subsequently interacts with the device end by voice, the language type set by the user can be directly obtained based on the user's voiceprint information and the preset voiceprint information.

[0096] In actual applications, the setting of the language type can be manifested as the device end can prompt the user with language type selection information, and at this time the device end can respond to the user's language type selection instruction, generate language type encoding information for the selected language type, and then store the language type encoding information of the selected language type and the voiceprint information encoding of the user's voiceprint information in the preset repository. That is, the voiceprint information encoding of the user's voiceprint information and the language type encoding information of the selected language type can be bound and stored in the preset repository, so as to achieve switching accuracy under multi-language switching based on the binding relationship and the voiceprint information encoding of the user's voiceprint information used to uniquely identify the user.

[0097] As an example, the prompt for the language type selection information can be implemented by voice prompting the language type selection information, such as a voice broadcast of "Please select the language type you want" on the device side; as another example, the prompt for the language type selection information can also be implemented by presenting a language type interface, which is not limited by the embodiments of the present invention.

[0098] It should be noted that there may be pre-established encoding rules for voiceprint information, and the encoding of voiceprint information may follow the established encoding rules. The encoding rules may involve data related to the voiceprint information, such as a combination of multiple feature values ​​including fundamental frequency, resonance peak, spectrum characteristics, pronunciation mode, timbre, etc. The embodiments of the present invention do not limit the specific encoding rules; and for language type encoding information, there may be unique language type encoding information corresponding to the language type, and the language type encoding information may be pre-set, that is, it is different from the generation of voiceprint information encoding, and the language type encoding information is not generated in real time. The embodiments of the present invention do not limit the pre-set specific language type encoding information.

[0099] Optionally, if no language selection instruction from the user is received within a preset time period, the preset language type is determined to be the language type selected for the user's voiceprint information.

[0100] Exemplarily, the user may be prompted to select a language through voice, for example, the device may make a voice announcement of "Please select the language you want" and wait to receive a language selection instruction issued by the user. After detecting the user's language selection instruction, such as "Sichuan dialect", the voiceprint information encoding of the user's voiceprint information and the language type encoding information corresponding to Sichuan dialect may be bound and stored in a preset repository; if the device is in a preset time period for issuing a voice announcement, for example, the waiting time for receiving instructions is 10s, if the user does not give a language selection instruction within 10s, the preset language may be selected by default, such as Mandarin, and the voiceprint information encoding of the user's voiceprint information and the language type encoding information corresponding to Mandarin may be bound and stored in the preset repository for voice interaction.

[0101] Optionally, when the voice recognition module is in the awake state, the device can also meet the user's demand for changing the language type.

[0102] When the user needs to change the language, the device can respond to the user's language switching instruction, obtain the language to be switched, and obtain the language selection information of the language to be switched, so as to bind the language encoding information corresponding to the language to be switched to the user voiceprint information of the user whose language is currently to be switched.

[0103] In one case, the user voiceprint information of the user whose language is to be switched may have language encoding information bound and stored in the preset storage repository. In this case, the language encoding information corresponding to the language to be switched can be overwritten and bound to the user voiceprint information of the user.

[0104] Specifically, when there is preset voiceprint information matching the user's voiceprint information in the preset storage repository, the voiceprint information code of the preset voiceprint information is obtained, and the language type selection information of the language to be switched is bound to the voiceprint information code of the preset voiceprint information and stored in the preset storage repository.

[0105] In another case, if the user voiceprint information of the user whose language is to be switched does not have language binding information in the preset storage library, the language encoding information corresponding to the language to be switched can be bound to the user voiceprint information of the user.

[0106] Specifically, when there is no preset voiceprint information matching the user's voiceprint information in the preset storage repository, a voiceprint information code for the user's voiceprint information is generated, and language type selection information is prompted. Then, in response to the user's language type selection instruction, language type coding information for the selected language type is generated, and the language type coding information for the selected language type and the voiceprint information code for the user's voiceprint information are stored in the preset storage repository.

[0107] As an example, the prompt for the language type selection information can be implemented by voice prompting the language type selection information, such as a voice broadcast of "Please select the language type you want" on the device side; as another example, the prompt for the language type selection information can also be implemented by presenting a language type interface, which is not limited by the embodiments of the present invention.

[0108] It should be noted that there may be pre-established encoding rules for voiceprint information, and the encoding of voiceprint information may follow the established encoding rules. The encoding rules may involve data related to the voiceprint information, such as a combination of multiple feature values ​​including fundamental frequency, resonance peak, spectrum characteristics, pronunciation mode, timbre, etc. The embodiments of the present invention do not limit the specific encoding rules; and for language type encoding information, there may be unique language type encoding information corresponding to the language type, and the language type encoding information may be pre-set, that is, it is different from the generation of voiceprint information encoding, and the language type encoding information is not generated in real time. The embodiments of the present invention do not limit the pre-set specific language type encoding information.

[0109] It should be noted that, in order to ensure accurate switching of the language types, in the embodiment of the present invention, a user has only one language type for interaction at the same time, and the language type can be switched by the user.

[0110] In the embodiment of the present invention, the user's voiceprint information is obtained by performing voiceprint recognition in response to a voice wake-up command, and the target language type encoding information of the preset voiceprint information is obtained according to the matching result of the user's voiceprint information and the preset voiceprint information, and the obtained target language type encoding information is transmitted to the cloud, so that the cloud can switch the language recognition model of the device to the language recognition model of the target language type based on the target language type encoding information, so that the device switched to the target language recognition model can perform voice broadcasting in the target language type. By adopting voice recognition technology and voiceprint recognition technology, the user's voice audio is analyzed and recognized to recognize the user's voiceprint information, and the target language type that the device needs to use is obtained, and then the target language type is automatically selected and used for voice interaction, which improves the switching accuracy in multiple language situations, thereby realizing flexible switching of multiple language types, meeting the interaction needs of different groups of people for language types, and can be applicable to voice interaction products in multiple language types.

[0111] Reference Figure 3 , shows a flowchart of another embodiment of a speech recognition method of a device of the present invention, which is applied to the cloud and may specifically include the following steps:

[0112] Step 301, the cloud receiving device transmits target language encoding information;

[0113] The device can be in sleep mode when not in use. If a user's voice wake-up command is detected, the device can enter the wake-up state from sleep mode. The device in the wake-up state can work normally and realize voice recognition.

[0114] In an embodiment of the present invention, the speech recognition performed by the device may be specifically manifested as recognition of the language type.

[0115] Among them, the voice wake-up instruction can be generated based on the detected user voice audio with a preset wake-up word. In one embodiment of the present invention, in order to identify the language type of the user who currently wakes up the device, the voiceprint recognition of the user who currently wakes up the device can be performed to obtain unique information for identifying the user and identify the language type of the user.

[0116] Specifically, for the user voiceprint information of a certain user, there may be a language type pre-set for the user's voiceprint information. At this time, after performing voiceprint recognition on the user who currently wakes up the device and obtaining the user's voiceprint information, the user's voiceprint information can be matched with the preset voiceprint information to obtain a matching result, and subsequent language type identification is performed based on the matching result.

[0117] The preset voiceprint information may be stored in a preset storage repository, and the preset voiceprint information in the preset storage repository has a preset language type by default.

[0118] Different voiceprint information is unique information used to identify different users, and voiceprint information can be used as a basis for switching languages.

[0119] In one case, if there is a preset voiceprint information matching the user's voiceprint information in the preset repository, it means that the user who currently wakes up the device has a preset language type, so that the device can recognize and respond to the instructions of the user who currently wakes up the device according to the set language type; in another case, if there is no preset voiceprint information matching the user's voiceprint information in the preset repository, it means that the user who currently wakes up the device does not have a preset language type. At this time, the language type can be set for the user voiceprint information of the user who currently wakes up the device and stored in the preset repository, so that when the user subsequently interacts with the device through voice, the language type set by the user can be obtained directly based on the user voiceprint information and the preset voiceprint information.

[0120] The information storage in the preset repository can be implemented based on the information storage module on the device side, which can be local storage. The language type set for a user's voiceprint information stored in the preset repository can be implemented based on the target language type encoding information of the voiceprint information.

[0121] In actual applications, the device's recognition of user commands can be manifested as identifying the language type of the corresponding voice commands through a language recognition model, and then identifying specific voice information. The language recognition model is deployed in the cloud. Since the voice recognition on the device side depends on the language recognition model, the device side can transmit the target language type encoding information to the cloud, and the cloud can receive the target language type encoding information transmitted by the device.

[0122] Step 302: Switching the language recognition model of the device to a target language recognition model of the target language based on the target language encoding information.

[0123] After receiving the target language encoding information transmitted by the device, the cloud can switch the language recognition model of the device to the target language recognition model of the target language based on the target language encoding information, thereby realizing flexible switching of multiple languages.

[0124] Switching the language recognition model of the device to the target language recognition model specifically means that the language recognition model currently used by the device is the target language recognition model, so as to switch the control device to the target language type, so that the device can perform voice recognition through the switched language recognition model, and realize voice interaction with the user using the language type.

[0125] In an embodiment of the present invention, by adopting speech recognition technology and voiceprint recognition technology, the user's voice audio is analyzed and identified to identify the user's voiceprint information, and the target language type that the device needs to use is obtained, and then the target language type is automatically selected and used for voice interaction, thereby improving the switching accuracy in multiple language situations, thereby achieving flexible switching of multiple language types, meeting the interaction needs of different groups of people for language types, and being applicable to voice interaction products in multiple language types; further, by adding the application of voiceprint technology, based on the uniqueness of the user's voiceprint information and the language type encoding information that is bound to the user's voiceprint information, the user's voiceprint information is used as the basis for identifying the language type, which can improve the switching accuracy in multiple language situations and thereby improve the user's voice interaction experience.

[0126] In one embodiment of the present invention, in order to facilitate those skilled in the art to understand the device voice recognition solution provided by the embodiment of the present invention, the following description is made in conjunction with a process schematic diagram:

[0127] Reference Figure 4 , which shows a schematic diagram of the process of voice recognition performed by the device provided in an embodiment of the present invention. Based on the implementation of the aforementioned process, it is possible to determine the type of language selected by the user.

[0128] Specifically, it can be performed as follows: S1 is executed to determine whether a voice wake-up word is detected. If a wake-up word is detected, the voice recognition module enters a wake-up state from a sleep mode. At this time, S2 can be executed to perform voiceprint recognition on the user voice audio containing the wake-up word, and the currently recognized user voiceprint information is compared / matched with the preset voiceprint information in the preset repository. S3 can be executed to determine whether there is preset voiceprint information matching the currently recognized user voiceprint information in the preset repository; if there is preset voiceprint information matching the currently recognized user voiceprint information in the preset repository, S7 can be executed to search the storage area for language type encoding information corresponding to this voiceprint information, and then S8 can be executed to transmit this language type encoding information to the cloud. S9: The cloud can switch to the corresponding language recognition model according to the language type encoding information. S10: The language type is switched successfully, and the device will use this language type for voice broadcasting.

[0129] If the wake-up word is not detected, the device can remain on standby.

[0130] If there is no preset voiceprint information matching the currently identified user voiceprint information in the preset storage repository, S4 can be executed at this time to save the voiceprint information and generate a voiceprint information code, which can be specifically manifested as generating a unique code for the currently identified user voiceprint information according to the established coding rules. Then S5 can be executed to enter the language selection and receive the user language selection instruction, which can be specifically manifested as prompting the user to select the language through voice, such as "Please select the language you want", and waiting to receive the user instruction. After the user gives the language selection instruction, such as "Sichuan dialect", S6 can be executed to bind the user voiceprint information code with the selected language code information for storage. The language code information here has been encoded and stored according to the protocol rules. When waiting to receive the user instruction, assuming that the waiting time for receiving the instruction is 10s, if the user does not give the language selection instruction within 10s, Mandarin can be used for voice interaction by default.

[0131] In addition, if the user needs to change the language type, he can give a language switching instruction after waking up, for example: "Switch language". At this time, the relevant steps mentioned above "If there is no preset voiceprint information in the preset repository that matches the currently identified user voiceprint information" can be repeated to overwrite and bind the language information and voiceprint information to the preset repository.

[0132] It should be noted that, for users who need to switch languages, their user voiceprint information may already have user voiceprint information encoding. When switching languages, the process of generating user voiceprint information encoding for the user voiceprint information can also be skipped. This is not limited in the embodiments of the present invention.

[0133] In an embodiment of the present invention, by adopting speech recognition technology and voiceprint recognition technology, the user's voice audio is analyzed and identified to identify the user's voiceprint information, and the target language type that the device needs to use is obtained, and then the target language type is automatically selected and used for voice interaction, thereby improving the switching accuracy in multiple language situations, thereby achieving flexible switching of multiple language types, meeting the interaction needs of different groups of people for language types, and being applicable to voice interaction products in multiple language types; further, by adding the application of voiceprint technology, based on the uniqueness of the user's voiceprint information and the language type encoding information that is bound to the user's voiceprint information, the user's voiceprint information is used as the basis for identifying the language type, which can improve the switching accuracy in multiple language situations and thereby improve the user's voice interaction experience.

[0134] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0135] Reference Figure 5 , shows a structural block diagram of a speech recognition device embodiment of a device of the present invention, which may specifically include the following modules:

[0136] The voiceprint recognition module 501 is used to respond to the voice wake-up command and perform voiceprint recognition to obtain the user's voiceprint information;

[0137] A voiceprint matching module 502, configured to obtain target language encoding information of the preset voiceprint information according to a matching result between the user voiceprint information and the preset voiceprint information;

[0138] A language recognition model switching module 503 is used to transmit the target language type encoding information to the cloud, and the cloud switches the language recognition model of the device to the target language type target language recognition model based on the target language type encoding information;

[0139] The voice playing module 504 is used for performing voice broadcasting according to the target language.

[0140] In an embodiment of the present invention, the preset voiceprint information is stored in a preset repository, in which a binding relationship between the voiceprint information encoding of the preset voiceprint information and the language type encoding information of the preset language type is pre-stored; the voiceprint matching module 502 may include the following submodules:

[0141] The target language encoding information acquisition submodule is used to respond to the user voiceprint information matching the preset voiceprint information and acquire the target language encoding information of the preset voiceprint information from the preset storage repository based on the binding relationship.

[0142] In an embodiment of the present invention, the voiceprint matching module 502 may further include the following submodules:

[0143] The voiceprint information storage submodule is used for storing the user voiceprint information in the preset storage repository in response to the absence of preset voiceprint information matching the user voiceprint information in the preset storage repository.

[0144] In one embodiment of the present invention, the voiceprint information storage submodule includes the following units:

[0145] The voiceprint information storage unit is used to generate a voiceprint information code for the user's voiceprint information and prompt language type selection information; in response to the user's language type selection instruction, generate language type coding information for the selected language type, and store the language type coding information of the selected language type and the voiceprint information code of the user's voiceprint information in the preset storage repository.

[0146] In one embodiment of the present invention, the voiceprint information storage unit may include the following subunits:

[0147] The voiceprint information storage subunit is used to bind the voiceprint information code of the user's voiceprint information and the language type code information of the selected language type and store them in the preset storage library.

[0148] In one embodiment of the present invention, the voiceprint information storage submodule may include the following units:

[0149] The language type setting unit is used to determine the preset language type as the language type selected for the user voiceprint information if no language type selection instruction from the user is received within a preset time period.

[0150] In an embodiment of the present invention, the voiceprint recognition module 501 may include the following submodules:

[0151] The voiceprint recognition submodule is used to detect the presence of a preset wake-up word in the user's voice audio, and switch the voice recognition module of the device from a sleep mode to a wake-up state; through the voice recognition module in the wake-up state, voiceprint recognition is performed on the user's voice audio to obtain the user's voiceprint information.

[0152] In an embodiment of the present invention, when the speech recognition module is in the awake state, the device provided by the embodiment of the present invention may further include the following modules:

[0153] A language switching module is used to respond to the user's language switching instruction, obtain the language type to be switched, and obtain the language type selection information of the language type to be switched; when there is preset voiceprint information matching the user's voiceprint information in the preset storage repository, obtain the voiceprint information code of the preset voiceprint information, bind the language type selection information of the language type to be switched with the voiceprint information code of the preset voiceprint information and store them in the preset storage repository.

[0154] In one embodiment of the present invention, the language switching module is also used to generate a voiceprint information code for the user's voiceprint information and prompt language type selection information when there is no preset voiceprint information matching the user's voiceprint information in the preset storage repository; in response to the user's language type selection instruction, generate language type coding information for the selected language type, and store the language type coding information of the selected language type and the voiceprint information code of the user's voiceprint information in the preset storage repository.

[0155] In an embodiment of the present invention, the voice recognition device of the device provided by the embodiment of the present invention obtains the user's voiceprint information by performing voiceprint recognition in response to a voice wake-up command, and obtains the target language type encoding information of the preset voiceprint information according to the matching result of the user's voiceprint information and the preset voiceprint information, and transmits the obtained target language type encoding information to the cloud, so that the cloud can switch the language recognition model of the device to the language recognition model of the target language type based on the target language type encoding information, so that the device switched to the target language recognition model can perform voice broadcasting according to the target language type. By adopting voice recognition technology and voiceprint recognition technology, the user's voice audio is analyzed and recognized to recognize the user's voiceprint information, and the target language type that the device needs to use is obtained, and then the target language type is automatically selected and used for voice interaction, so as to improve the switching accuracy in multiple language situations, thereby realizing flexible switching of multiple language types, meeting the interaction needs of different groups of people for language types, and being applicable to voice interaction products in multiple language types.

[0156] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0157] An embodiment of the present invention further provides an electronic device, including:

[0158] It includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, the various processes of the speech recognition method embodiment of the above-mentioned device are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0159] An embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the speech recognition method embodiment of the above-mentioned device are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0160] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0161] It will be appreciated by those skilled in the art that the embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0162] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0163] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0164] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0165] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0166] Finally, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0167] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0168] The speech recognition method of a device, a speech recognition device of a device, a corresponding electronic device and a corresponding computer-readable storage medium provided by the present invention are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A speech recognition method for a device, characterized in that: The method comprises: Respond to the voice wake-up command and perform voiceprint recognition to obtain the user's voiceprint information; According to the matching result between the user voiceprint information and the preset voiceprint information, obtaining the target language type encoding information of the preset voiceprint information; Transmitting the target language encoding information to the cloud, and the cloud switching the language recognition model of the device to the target language recognition model of the target language based on the target language encoding information; The voice broadcast is performed according to the target language.

2. The method according to claim 1, characterized in that The preset voiceprint information is stored in a preset storage repository, in which a binding relationship between the voiceprint information code of the preset voiceprint information and the language type code information of the preset language type is pre-stored; The step of obtaining target language encoding information of the preset voiceprint information according to a matching result between the user voiceprint information and the preset voiceprint information includes: In response to the user voiceprint information matching the preset voiceprint information, based on the binding relationship, the target language type encoding information of the preset voiceprint information is obtained from the preset storage repository.

3. The method according to claim 1 or 2, characterized in that: The step of obtaining target language encoding information of the preset voiceprint information in response to the matching result between the user voiceprint information and the preset voiceprint information further includes: In response to the absence of preset voiceprint information matching the user voiceprint information in the preset storage repository, the user voiceprint information is stored in the preset storage repository.

4. The method according to claim 3, characterized in that The storing the user voiceprint information into the preset storage repository includes: Generate a voiceprint information code for the user's voiceprint information and prompt language selection information; In response to the user's language selection instruction, language type coding information for the selected language type is generated, and the language type coding information of the selected language type and the voiceprint information coding of the user's voiceprint information are stored in the preset storage repository.

5. The method according to claim 1, characterized in that The responding to the voice wake-up instruction and performing voiceprint recognition to obtain the user's voiceprint information includes: Detecting the presence of a preset wake-up word in the user's voice audio, switching the voice recognition module of the device from a sleep mode to a wake-up state; The voiceprint recognition is performed on the user's voice audio through the voice recognition module in the awake state to obtain the user's voiceprint information.

6. The method according to claim 5, characterized in that When the speech recognition module is in the awake state, the method further includes: In response to a user's language switching instruction, obtain the language type to be switched, and obtain language type selection information of the language type to be switched; When there is preset voiceprint information matching the user's voiceprint information in the preset storage repository, the voiceprint information code of the preset voiceprint information is obtained, and the language type selection information of the language to be switched is bound to the voiceprint information code of the preset voiceprint information and stored in the preset storage repository.

7. The method according to claim 5, characterized in that The method further comprises: When there is no preset voiceprint information matching the user's voiceprint information in the preset storage repository, generating a voiceprint information code for the user's voiceprint information, and prompting language selection information; In response to the user's language selection instruction, language type coding information for the selected language type is generated, and the language type coding information of the selected language type and the voiceprint information coding of the user's voiceprint information are stored in the preset storage repository.

8. The method according to claim 4 or 7, characterized in that: The step of storing the language type encoding information of the selected language type and the voiceprint information encoding of the user voiceprint information in the preset storage repository includes: The voiceprint information encoding of the user's voiceprint information and the language type encoding information of the selected language type are bound and stored in the preset storage repository.

9. The method according to claim 4, characterized in that The method further comprises: If no language selection instruction from the user is received within a preset time period, the preset language type is determined to be the language type selected for the user's voiceprint information.

10. A speech recognition device for a device, characterized in that: The device comprises: The voiceprint recognition module is used to respond to the voice wake-up command and perform voiceprint recognition to obtain the user's voiceprint information; A voiceprint matching module, configured to obtain target language encoding information of the preset voiceprint information according to a matching result between the user voiceprint information and the preset voiceprint information; A language recognition model switching module, configured to transmit the target language encoding information to the cloud, and the cloud switches the language recognition model of the device to a target language recognition model of the target language based on the target language encoding information; The voice playing module is used for performing voice broadcasting in the target language.

11. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein when the computer program is executed by the processor, the speech recognition method of the device according to any one of claims 1 to 9 is implemented.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the speech recognition method of the device according to any one of claims 1 to 9 is implemented.

Images