Voiceprint registration method, device and computer-readable storage medium

Abstract

The present invention discloses a voiceprint registration method, device and computer-readable storage medium, and relates to the field of speech technology. The voiceprint registration method includes: generating guidance information according to a reading text and a process text for guiding a user to speak the reading text with a preset sound attribute; after the guidance information is output to the user, obtaining the user's reading voice; determining the sound attribute of the reading voice, and identifying the text content corresponding to the reading voice; determining that the reading voice is available when the sound attribute of the reading voice is a preset sound attribute and the text content corresponding to the reading voice matches the reading text; for the guidance information used for voiceprint registration, storing the corresponding relationship between the user and the voiceprint information extracted from the reading voice when the reading voice is available. The present invention improves the applicable population and convenience of voiceprint registration, and also improves the efficiency of the registration process.

Description

Technical Field

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

[0002] Voiceprint recognition technology determines who the current audio comes from based on a certain audio. It needs to have prior information such as a speaker model. In voiceprint recognition, it is necessary to obtain information about a specific person or multiple people in advance, and then when an unknown audio is obtained, it is necessary to determine whether the audio belongs to one of the existing speakers.

[0003] The basic steps of voiceprint recognition can include the following: first, use a large number of speakers' voices to train a voiceprint recognition model; then registration, that is, the device needs to first record the audio of a speaker to generate a speaker model of the speaker; then testing, that is, matching the unknown test audio with the speaker model to determine whether the test audio belongs to the specified speaker. Voiceprint recognition is different from speech recognition. Voiceprint recognition does not need to judge the corresponding text information based on the audio, and does not need to determine the speaker's age, gender and other prior information.

[0004] Voiceprint recognition can be applied to daily smart devices to provide personalized services, and can also be used in the financial and security fields to confirm identities. This requires voiceprint recognition to have relatively high accuracy and relatively high anti-attack capabilities.

[0005] In the related art, during the registration phase of voiceprint recognition, the speaker enters the audio corresponding to the text information according to the text information prompted by the device, and the recorded audio is greater than or equal to one. Summary of the invention

[0006] After analyzing the relevant technologies, the inventors found that in the relevant technologies, voiceprint recognition only requires users to speak the audio corresponding to the text, but some users use non-Mandarin, which may lead to inaccurate recognition or an unsmooth registration process.

[0007] A technical problem to be solved by the embodiments of the present invention is: how to improve the applicability of voiceprint registration.

[0008] According to a first aspect of some embodiments of the present invention, a voiceprint registration method is provided, comprising: generating guidance information based on a read text and a process text for guiding a user to speak the read text using preset sound attributes; obtaining the user's read voice after the guidance information is output to the user; determining the sound attributes of the read voice, and identifying the text content corresponding to the read voice; determining that the read voice is available when the sound attributes of the read voice are preset sound attributes and the text content corresponding to the read voice matches the read text; for the guidance information used for voiceprint registration, storing the correspondence between the user and the voiceprint information extracted from the read voice when the read voice is available.

[0009] In some embodiments, the guidance information is a guidance voice.

[0010] In some embodiments, a guidance voice with preset sound attributes is generated based on the read text and the process text.

[0011] In some embodiments, when the sound attribute of the read-aloud voice is a preset sound attribute, the text content corresponding to the read-aloud voice matches the read-aloud text, and the read-aloud voice is not noise, it is determined that the read-aloud voice is available.

[0012] In some embodiments, the voiceprint registration method also includes: determining that the reading voice is unavailable when the sound attribute of the reading voice is not a preset sound attribute, or the text content corresponding to the reading voice does not match the reading text, or the reading voice is noise; when the reading voice is unavailable, outputting corresponding reading correction information according to the unavailable type of the reading voice, wherein the unavailable type includes mismatch of sound attributes, incomplete content of the reading voice, and noise of the reading voice.

[0013] In some embodiments, when the text content corresponding to the read voice includes the read text, or when the phoneme sequence corresponding to the text content corresponding to the read voice includes the phoneme sequence corresponding to the read text, the text content corresponding to the read voice matches the read text.

[0014] In some embodiments, the voiceprint registration method further includes: obtaining sound attributes in the registration information input by the user as preset sound attributes; or, before generating the guidance information, collecting the user's voice and determining the sound attributes of the collected user's voice as preset sound attributes.

[0015] In some embodiments, determining the sound attribute of the read aloud speech includes: inputting the speech features of the read aloud speech into a preset sound attribute classification model to obtain the sound attribute of the read aloud speech.

[0016] In some embodiments, determining the sound attributes of the read aloud speech includes: inputting the speech features of the read aloud speech into a preset neural network model to obtain a speech embedding feature vector extracted by a hidden layer of the neural network model; calculating the distance between the speech embedding feature vector and the preset speech embedding feature vector of each sound attribute, and determining the shortest distance therebetween; when the shortest distance is not greater than a preset distance threshold, determining the sound attribute corresponding to the shortest distance as the sound attribute of the read aloud speech; when the shortest distance is greater than the preset distance threshold, determining the sound attribute of the read aloud speech as an unknown attribute.

[0017] In some embodiments, identifying text content corresponding to the read-aloud speech includes: using a speech recognition model corresponding to the sound attributes of the read-aloud speech to determine the text content corresponding to the read-aloud speech.

[0018] In some embodiments, the voiceprint registration method further includes: randomly selecting a preset number of characters from a character library to form an alternative text, and determining a phoneme combination of the alternative text; detecting the frequency of occurrence of the phoneme combination in a preset text library; and when the frequency of occurrence of the phoneme combination is lower than the preset frequency, using the alternative text as the text to be read aloud.

[0019] In some embodiments, an isolated word recognition model corresponding to a preset sound attribute is used to recognize the read speech, and the text content corresponding to the read speech is obtained.

[0020] In some embodiments, the guidance information is a guidance voice, and the process text includes a description text corresponding to each randomly selected word.

[0021] In some embodiments, the voice attribute is a dialect type.

[0022] In some embodiments, for guidance information for voiceprint verification, when the read voice is available, if the extracted voiceprint matches the stored voiceprint corresponding to the user, the user's identity verification is passed.

[0023] According to a second aspect of some embodiments of the present invention, there is provided a voiceprint registration device, comprising: a guidance information generation module, configured to generate guidance information based on a reading text and a process text for guiding a user to speak the reading text with preset sound attributes; a reading voice acquisition module, configured to acquire the user's reading voice after the guidance information is output to the user; a reading voice analysis module, configured to determine the sound attributes of the reading voice and identify the text content corresponding to the reading voice; an availability determination module, configured to determine that the reading voice is available when the sound attributes of the reading voice are preset sound attributes and the text content corresponding to the reading voice matches the reading text; and a storage module, configured to store, for the guidance information used for voiceprint registration, the correspondence between the user and the voiceprint information extracted from the reading voice when the reading voice is available.

[0024] According to a third aspect of some embodiments of the present invention, there is provided a voiceprint registration device, comprising: a memory; and a processor coupled to the memory, wherein the processor is configured to execute any of the aforementioned voiceprint registration methods based on instructions stored in the memory.

[0025] According to a fourth aspect of some embodiments of the present invention, there is provided a voiceprint registration system, comprising: any one of the aforementioned voiceprint registration devices; an output device configured to output guidance information generated by the voiceprint registration device; and a recording device configured to record the user's reading voice.

[0026] In some embodiments, the output device is a sound output device.

[0027] According to a fifth aspect of some embodiments of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, any of the aforementioned voiceprint registration methods is implemented.

[0028] Some embodiments of the above invention have the following advantages or beneficial effects: the embodiments of the present invention can guide the user to speak the preset reading text with preset voice attributes through automatically generated guidance information. In addition, verification and voice collection can be performed according to the user's speaking habits, and voice recognition can be performed when the voice attributes used by the user are understood, thereby improving the accuracy of recognition. Therefore, the applicable population and convenience of voiceprint registration are improved, and the efficiency of the registration process is also improved.

[0029] Further features and advantages of the present invention will become apparent from the following detailed description of exemplary embodiments of the present invention with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0031] Figure 1A A schematic flow chart of a voiceprint registration method according to some embodiments of the present invention is shown.

[0032] Figure 1B A schematic flow chart of a voiceprint verification method according to some embodiments of the present invention is shown.

[0033] Figure 2 A schematic flow chart of a method for outputting guidance speech according to some embodiments of the present invention is shown.

[0034] Figure 3 A schematic flow chart of a method for correcting reading aloud according to some embodiments of the present invention is shown.

[0035] Figure 4 A schematic flow chart of a method for generating a read-aloud text according to some embodiments of the present invention is shown.

[0036] Figure 5 A schematic diagram of the structure of a voiceprint registration device according to some embodiments of the present invention is shown.

[0037] Figure 6 A schematic structural diagram of a voiceprint registration system according to some embodiments of the present invention is shown.

[0038] Figure 7 A schematic diagram of the structure of a voiceprint registration device according to some other embodiments of the present invention is shown.

[0039] Figure 8 A schematic structural diagram of a voiceprint registration device according to some further embodiments of the present invention is shown. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. The following description of at least one exemplary embodiment is actually only illustrative and is by no means intended to limit the present invention and its application or use. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] The relative arrangement of components and steps, the numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.

[0042] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0043] Technologies, methods, and apparatus known to ordinary technicians in the relevant field may not be discussed in detail, but where appropriate, such technologies, methods, and apparatus should be considered part of the authorization specification.

[0044] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0045] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0046] Figure 1A FIG. 2 is a flow chart showing a voiceprint registration method according to some embodiments of the present invention. Figure 1A As shown, the voiceprint registration method of this embodiment includes steps S102 to S110.

[0047] In step S102, guidance information is generated according to the reading text and the process text for guiding the user to speak the reading text with preset sound attributes.

[0048] In some embodiments, the voice attribute refers to the dialect type. When the user can only speak a dialect and cannot speak Mandarin, the user is guided to speak the preset content in the dialect; or, when the user can speak Mandarin, the user is guided to speak the preset content in Mandarin. Thus, for different dialects under the same language, a personalized voiceprint registration method can also be provided, which improves the applicability of voiceprint registration and the convenience of user use.

[0049] In some embodiments, the sound attributes in the registration information input by the user are obtained as preset sound attributes.

[0050] In some embodiments, before generating the guidance information, the user's voice is collected, and the sound attributes of the collected user's voice are determined as preset sound attributes. For example, before generating the guidance information, guidance information such as "Please say a few words" is played or displayed to allow the user to say something casually in a relaxed state, and the sound attributes used by the user are identified. Thus, the sound attributes that the user is more accustomed to using can be detected more accurately.

[0051] The process text includes some guiding words, such as "Please say it", "Please repeat it", etc. In some embodiments, the process text includes a text synthesized by using a description text with preset sound attributes and a preset guiding template. For example, the description text of Cantonese is "Cantonese", and the guiding template is "Please use <description text> to say it", where "<>" is a placeholder for the description text to replace. The synthesized text is "Please use Cantonese to say it".

[0052] In some embodiments, the guidance information is used for voiceprint registration or voiceprint verification. Each type of guidance information corresponds to a preset process text, so that the user can clearly know whether the current process is a registration process or a verification process.

[0053] In step S104, after the guide information is output to the user, the user's reading voice is acquired.

[0054] An exemplary process of guiding the user to speak the read text is as follows. For a user who speaks Sichuan dialect, when the user registers, the read text is "Good morning", and the process text is, for example, "Please speak in Sichuan dialect". The content played or displayed on the user's terminal is "Please speak in Sichuan dialect for good morning", so that after hearing the played content, the user will say "Good morning" in Sichuan dialect.

[0055] In step S106 , the sound attributes of the read-aloud voice are determined, and the text content corresponding to the read-aloud voice is recognized.

[0056] In some embodiments, the speech features of the read speech are input into a preset sound attribute classification model to obtain the sound attributes of the read speech. The sound attribute classification model is, for example, a deep neural network model trained by using labeled sound samples. The sound attribute classification model can identify preset types of sound attributes and identify any input as one of the preset types.

[0057] In some embodiments, the speech features of the speech read aloud are input into a preset neural network model to obtain a speech embedding feature vector extracted by the hidden layer of the neural network model, for example, extracting a feature vector output by one or two layers after the pooling layer; calculating the distance between the speech embedding feature vector and the preset speech embedding feature vector of each sound attribute, and determining the shortest distance therebetween; when the shortest distance is not greater than a preset distance threshold, determining the sound attribute corresponding to the shortest distance as the sound attribute of the speech read aloud; when the shortest distance is greater than a preset distance threshold, determining the sound attribute of the speech read aloud as an unknown attribute. In this way, when the sound attribute used by the user does not belong to any of the preset sound attributes, it is determined as an unknown attribute.

[0058] The above two sound attribute classification methods can also be applied to the recognition of other voices. For example, before generating the guidance information, the user's voice is collected, and the sound attributes of the collected user's voice are determined by one of the above two methods.

[0059] In some embodiments, the speech recognition model corresponding to the sound attribute of the read speech is used to determine the text content corresponding to the read speech. For example, if the user speaks the read speech in Mandarin, the speech recognition model corresponding to Mandarin or a general speech recognition model is used; if the user speaks the read speech in Sichuan dialect, the speech recognition model corresponding to Sichuan dialect is used. Thus, the user's speech content can be recognized more accurately.

[0060] In step S108 , when the sound attribute of the read-aloud voice is a preset sound attribute and the text content corresponding to the read-aloud voice matches the read-aloud text, it is determined that the read-aloud voice is available.

[0061] In some embodiments, when the text content corresponding to the read voice includes the read text, the text content corresponding to the read voice matches the read text.

[0062] In some embodiments, when the phoneme sequence corresponding to the text content corresponding to the read speech includes the phoneme sequence corresponding to the read text, the text content corresponding to the read speech matches the read text. For example, the words in the read text include "extremely", and the corresponding content in the recognition result of the user's read speech is "jealous". Although the text in the recognition result is inconsistent with the read text, the pronunciation is consistent. In this case, it can also be considered that the text content corresponding to the read speech matches the read text.

[0063] During the voiceprint registration phase, the main purpose of speech recognition is to detect liveness and fully collect the user's pronunciation of various phonemes, and the accuracy of text information is not required. Therefore, the method of confirming text matching through phoneme matching can improve registration efficiency.

[0064] By confirming that the sound attribute of the read voice is the preset sound attribute, the registered sound attribute can be made consistent with the sound attribute that the user is accustomed to using or the sound attribute selected by the user. When each sound attribute corresponds to a speech recognition model, the recognition accuracy can be improved.

[0065] By confirming whether the text content corresponding to the read voice matches the read text, liveness detection can be achieved, reducing the possibility that the speaker is not a real person but a recording. This improves the security of registration. In addition, when the read text includes many phonemes, the method of determining whether the read voice is available based on the speech recognition results can also make the user's voice cover more phonemes to generate a more comprehensive speaker model.

[0066] In some embodiments, when the sound attribute of the read speech is a preset sound attribute, the text content corresponding to the read speech matches the read text, and the read speech is not noise, it is determined that the read speech is available. For example, a Voice Activity Detection (VAD) model is used to identify whether the collected sound is speech or noise.

[0067] In step S110, for the guidance information for voiceprint registration, when the reading voice is available, the corresponding relationship between the user and the voiceprint information extracted from the reading voice is stored through registration.

[0068] Through the above embodiments, the user can be guided to speak the preset reading text with the preset voice attributes through the automatically generated guidance information. In addition, verification and voice collection can be performed according to the user's speaking habits, and voice recognition can be performed when the voice attributes used by the user are understood, thereby improving the recognition accuracy. Therefore, the above embodiments increase the applicable population and convenience of voiceprint registration, and also improve the efficiency of the registration process.

[0069] In some embodiments, after completing registration, when the user subsequently uses the relevant application or product, the user speaks with the preset voice attributes. By extracting the voiceprint in the user's voice and comparing it with the voiceprint stored during the registration phase, it can be determined whether the speaker is a registered user. Figure 1B An embodiment of the voiceprint verification method of the present invention is described.

[0070] Figure 1B FIG. 2 is a flow chart showing a voiceprint verification method according to some embodiments of the present invention. Figure 1B As shown, the voiceprint verification method of this embodiment includes steps S112 to S120, wherein the specific implementation of steps S112 to S118 refers to steps S102 to S108 in the embodiment of FIG. 1 , which will not be repeated here.

[0071] In step S120, for the guidance information for voiceprint verification, when the reading voice is available, if the extracted voiceprint matches the stored voiceprint corresponding to the user, the identity verification of the user is passed.

[0072] Through the above embodiments, in the voiceprint verification stage, the user can be guided to speak the preset reading text with the preset voice attributes through the automatically generated guidance information. In addition, verification and voice collection can be performed according to the user's speaking habits, and voice recognition can be performed when the voice attributes used by the user are understood, thereby improving the recognition accuracy. Therefore, the above embodiments improve the applicable population and convenience of the voiceprint-based verification process, and also improve the efficiency of the verification process.

[0073] In some embodiments, the guidance information is a guidance voice, thereby prompting the user to speak the preset content using corresponding sound attributes in a voice manner. Figure 2 An embodiment of a guidance speech output method is described.

[0074] Figure 2 FIG. 2 is a flow chart showing a method for outputting guidance speech according to some embodiments of the present invention. Figure 2 As shown, the guidance voice output method of this embodiment includes steps S202 to S206.

[0075] In step S202, a guide text is generated according to the reading text and a process text for guiding the user to speak the reading text with a preset sound attribute.

[0076] In step S204, the guidance text is converted into guidance speech.

[0077] In step S206, the guidance voice is played to the user.

[0078] In some embodiments, the guidance text is converted into guidance voice using Text To Speech (TTS) technology. In some embodiments, the generated guidance voice is a voice with certain emotions and intonations, thereby enhancing the fun and interactivity of the voice guidance process. In some embodiments, the generated guidance voice is a voice with preset sound attributes, thereby making it more helpful for users using the corresponding sound attributes to understand what is to be said and to be able to speak more accurately using the corresponding sound attributes.

[0079] By using voice to play guidance information, users with reading disabilities such as illiteracy, incomplete literacy, poor eyesight or blindness can also obtain the information to be read aloud, thereby increasing the applicability of voiceprint recognition applications.

[0080] In some embodiments, when it is recognized that the user's reading voice does not meet the requirements, corresponding correction information can be given according to the specific situation to guide the user to speak the preset content more accurately. Figure 3 An embodiment of the reading correction method of the present invention is described.

[0081] Figure 3 FIG. 2 is a flow chart showing a method for correcting aloud reading according to some embodiments of the present invention. Figure 3 As shown, the reading correction method of this embodiment includes steps S302 to S312.

[0082] In step S302, after the guidance information is output to the user, the user's reading voice is obtained. The generation method of the guidance information is as described in the above embodiment, which will not be described again here.

[0083] In step S304, the sound attribute of the read speech is determined, the text content corresponding to the read speech is recognized, and it is detected whether the read speech is noise.

[0084] In step S306, when the sound attribute of the read voice is a preset sound attribute, the text content corresponding to the read voice matches the read text, and the read voice is not noise, it is determined that the read voice is available.

[0085] In step S308, the correspondence between the user and the voiceprint information extracted from the read speech is stored.

[0086] In step S310 , when the sound attribute of the read voice is not a preset sound attribute, or the text content corresponding to the read voice does not match the read text, or the read voice is noise, it is determined that the read voice is unavailable.

[0087] In step S312, when the reading voice is unavailable, the corresponding reading correction information is output as guidance information according to the unavailable type of the reading voice, wherein the unavailable type includes mismatch of sound attributes, incomplete content of the reading voice, and noise of the reading voice. Then, the process returns to step S302 to continue to obtain the user's reading voice.

[0088] For example, when the preset dialect type is Mandarin, and the user speaks the reading voice in Sichuan dialect, the reading correction information is "Please repeat in Mandarin" and similar information; when the user only speaks part of the reading text, the reading correction information is "The content is missing, please say it again in full" and similar information; when the reading voice is noise, the reading correction information is "It is not clear, please go to a quiet place and say it again." Each type of correction information can also include information corresponding to the reading text. The reading correction information is output to the user through text display or voice playback.

[0089] Through the method of the above embodiment, guidance and correction information can be automatically generated during voiceprint application processes such as user voiceprint registration and recognition to assist users in completing voiceprint application-related operations more quickly and accurately.

[0090] In some embodiments, when generating a text to be read aloud, words, phrases or sentences with a frequency lower than a preset value are selected to reduce the possibility of an attacker recording in advance to attack. In some embodiments, meaningless "self-made words" can also be automatically generated to further reduce the possibility of being attacked. Figure 4 An embodiment of a method for generating a spoken text is described.

[0091] Figure 4 FIG. 2 is a flow chart showing a method for generating a text read aloud according to some embodiments of the present invention. Figure 4 As shown in Figure 4 , the method for generating the text to be read aloud in this embodiment includes steps S402 to S406.

[0092] In step S402, a preset number of characters are randomly selected from the character library to form an alternative text, and the phoneme combination of the alternative text is determined.

[0093] In step S404, the occurrence frequency of the phoneme combination is detected in the preset text library.

[0094] In step S406, when the occurrence frequency of the phoneme combination is lower than the preset frequency, the alternative text is used as the text to be read aloud.

[0095] For example, when randomly selecting "ming" and "tian" to form an alternative text, although there is no such word as "mingtian" in the existing word library, this word and the common word "tomorrow" include exactly the same phonemes, so the occurrence frequency of the phoneme combination of this alternative text is relatively high.

[0096] In some embodiments, when the guiding information is in the form of guiding speech, the description text of each randomly selected character is found and used as part of the process text, so as to convert the description text into voice for playback. For example, if the coined word is "jiaoqi", then "jiao means relatively" and "qi means strange" can be used as the description texts of the two characters in the coined word respectively. The content of the guiding speech is, for example, "Please say jiaoqi, jiao means relatively, qi means strange". In this way, it can assist the user to more vividly understand the content to be read aloud, so that the user can more accurately say the preset content.

[0097] In some embodiments, when using a coined word as the text to be read aloud and recognizing the user's read-aloud speech, an isolated-word recognition model corresponding to the preset sound attributes is used to recognize the read-aloud speech, and the text content corresponding to the read-aloud speech is obtained. Generally, the speech recognition model contains a multi-language model, such as 2-Gram (bigram) and 3-Gram (trigram) models. When using a coined word, an isolated-word recognition model can be used during the recognition process. This model, for example, uses a 1-Gram (unigram) model or does not use a grammar model, so as to only consider the speech aspect during the recognition process and not consider the influence of semantics on the recognition result.

[0098] Through the method of the above embodiments, text with low arousal can be automatically synthesized, thereby improving safety.

[0099] Next, refer to Figure 5 the embodiments of the voiceprint registration device are described.

[0100] Figure 5 The structural schematic diagram of the voiceprint registration device according to some embodiments of the present invention is shown. As Figure 5As shown, the voiceprint registration device 500 of this embodiment includes: a guidance information generation module 5100, configured to generate guidance information according to a reading text and a process text for guiding a user to speak the reading text with a preset sound attribute; a reading voice acquisition module 5200, configured to acquire the user's reading voice after the guidance information is output to the user; a reading voice analysis module 5300, configured to determine the sound attribute of the reading voice and identify the text content corresponding to the reading voice; an availability determination module 5400, configured to determine that the reading voice is available when the sound attribute of the reading voice is the preset sound attribute and the text content corresponding to the reading voice matches the reading text; a storage module 5500, configured to store the corresponding relationship between the user and the voiceprint information extracted from the reading voice for the guidance information used for voiceprint registration when the reading voice is available.

[0101] In some embodiments, the guidance information is a guidance voice.

[0102] In some embodiments, the guidance information generation module 5100 is further configured to generate a guidance voice with preset sound attributes based on the read text and the process text.

[0103] In some embodiments, the availability determination module 5400 is further configured to determine that the reading voice is available when the sound attribute of the reading voice is a preset sound attribute, the text content corresponding to the reading voice matches the reading text, and the reading voice is not noise.

[0104] In some embodiments, the availability determination module 5400 is further configured to determine that the reading voice is unavailable when the sound attributes of the reading voice are not preset sound attributes, or the text content corresponding to the reading voice does not match the reading text, or the reading voice is noise; the guidance information generation module 5100 is further configured to generate corresponding reading correction information and add it to the guidance information according to the unavailability type of the reading voice when the reading voice is unavailable, wherein the unavailability type includes mismatch of sound attributes, incomplete content of the reading voice, and noise of the reading voice.

[0105] In some embodiments, when the text content corresponding to the read voice includes the read text, or when the phoneme sequence corresponding to the text content corresponding to the read voice includes the phoneme sequence corresponding to the read text, the text content corresponding to the read voice matches the read text.

[0106] In some embodiments, the voiceprint registration device 500 also includes: a sound attribute acquisition module 5600, which is configured to obtain the sound attributes in the registration information input by the user as preset sound attributes; or, before generating the guidance information, collect the user's voice and determine the sound attributes of the collected user's voice as preset sound attributes.

[0107] In some embodiments, the reading speech analysis module 5300 is further configured to input the speech features of the reading speech into a preset sound attribute classification model to obtain the sound attributes of the reading speech.

[0108] In some embodiments, the reading speech analysis module 5300 is further configured to input the speech features of the reading speech into a preset neural network model to obtain a speech embedding feature vector extracted by the hidden layer of the neural network model; calculate the distance between the speech embedding feature vector and the preset speech embedding feature vector of each sound attribute, and determine the shortest distance therebetween; when the shortest distance is not greater than a preset distance threshold, determine the sound attribute corresponding to the shortest distance as the sound attribute of the reading speech; when the shortest distance is greater than the preset distance threshold, determine the sound attribute of the reading speech as an unknown attribute.

[0109] In some embodiments, the reading speech analysis module 5300 is further configured to use a speech recognition model corresponding to the sound attributes of the reading speech to determine the text content corresponding to the reading speech.

[0110] In some embodiments, the voiceprint registration device 500 also includes: a reading text generation module 5700, which is configured to randomly select a preset number of characters from the character library to form an alternative text, and determine the phoneme combination of the alternative text; detect the frequency of occurrence of the phoneme combination in the preset text library; when the frequency of occurrence of the phoneme combination is lower than the preset frequency, use the alternative text as the reading text.

[0111] In some embodiments, the reading speech analysis module 5300 is further configured to recognize the reading speech using an isolated word recognition model corresponding to a preset sound attribute, and obtain text content corresponding to the reading speech.

[0112] In some embodiments, the guidance information is a guidance voice, and the process text includes a description text corresponding to each randomly selected word.

[0113] In some embodiments, the voice attribute is a dialect type.

[0114] In some embodiments, the voiceprint registration device 500 further includes a verification module 5800 configured to verify the user's identity if the extracted voiceprint matches the stored voiceprint corresponding to the user, for guidance information for voiceprint verification, when the reading voice is available.

[0115] Reference below Figure 6 An embodiment of the voiceprint registration system of the present invention is described.

[0116] Figure 6 FIG. 2 shows a schematic diagram of the structure of a voiceprint registration system according to some embodiments of the present invention. Figure 6 As shown, the voiceprint registration system 60 of this embodiment includes: a voiceprint registration device 610, whose specific real-time method can refer to the voiceprint registration device 500; an output device 620, configured to output the guidance information generated by the voiceprint registration device; and a recording device 630, configured to record the user's reading voice.

[0117] In some embodiments, output device 620 is a sound output device.

[0118] Figure 7 FIG. 2 shows a schematic diagram of the structure of a voiceprint registration device according to some other embodiments of the present invention. Figure 7 As shown, the voiceprint registration device 70 of this embodiment includes: a memory 710 and a processor 720 coupled to the memory 710 , and the processor 720 is configured to execute the voiceprint registration method in any one of the aforementioned embodiments based on the instructions stored in the memory 710 .

[0119] The memory 710 may include, for example, a system memory, a fixed non-volatile storage medium, etc. The system memory may store, for example, an operating system, an application program, a boot loader, and other programs.

[0120] Figure 8 FIG. 2 shows a schematic diagram of the structure of a voiceprint registration device according to some other embodiments of the present invention. Figure 8 As shown, the voiceprint registration device 80 of this embodiment includes: a memory 810 and a processor 820, and may also include an input / output interface 830, a network interface 840, a storage interface 850, etc. These interfaces 830, 840, 850 and the memory 810 and the processor 820 may be connected, for example, via a bus 860. Among them, the input / output interface 830 provides a connection interface for input / output devices such as a display, a mouse, a keyboard, and a touch screen. The network interface 840 provides a connection interface for various networked devices. The storage interface 850 provides a connection interface for external storage devices such as SD cards and USB flash drives.

[0121] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the program implements any of the aforementioned voiceprint registration methods when executed by a processor.

[0122] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, 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 present invention may take the form of a computer program product implemented on one or more computer-usable non-transient storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0123] The present invention is described with reference to the flowchart and / or block diagram of the method, device (system), and computer program product according to the embodiment of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0124] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0125] These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0126] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A voiceprint registration method, include: Randomly selecting a preset number of characters from a character library to form candidate texts, and determining phoneme combinations of the candidate texts; Detecting the occurrence frequency of the phoneme combination in a preset text library; When the occurrence frequency of the phoneme combination is lower than a preset frequency, using the candidate text as the text to be read aloud; Generating guidance information according to the read text and a process text for guiding the user to speak the read text with a preset sound attribute, wherein the preset sound attribute is a sound attribute in the user's registration information or a sound attribute of the user's voice collected before generating the guidance information; After the guidance information is output to the user, obtaining the user's reading voice; Determining the sound attributes of the read voice and identifying the text content corresponding to the read voice; In the case where the sound attribute of the read-aloud voice is the preset sound attribute and the text content corresponding to the read-aloud voice matches the read-aloud text, determining that the read-aloud voice is available, wherein in the case where the phoneme sequence corresponding to the text content corresponding to the read-aloud voice includes the phoneme sequence corresponding to the read-aloud text, the text content corresponding to the read-aloud voice matches the read-aloud text; For the guidance information for voiceprint registration, when the reading voice is available, the corresponding relationship between the user and the voiceprint information extracted from the reading voice is stored.

2. The voiceprint registration method according to claim 1, in, The guidance information is a guidance voice.

3. The voiceprint registration method according to claim 2, in, A guiding voice with the preset sound attribute is generated according to the reading text and the process text.

4. The voiceprint registration method according to claim 1, In the case where the sound attribute of the read aloud voice is the preset sound attribute, the text content corresponding to the read aloud voice matches the read aloud text, and the read aloud voice is not noise, it is determined that the read aloud voice is available.

5. The voiceprint registration method according to claim 1, further comprising: include: In the case where the sound attribute of the read-aloud voice is not the preset sound attribute, or the text content corresponding to the read-aloud voice does not match the read-aloud text, or the read-aloud voice is noise, determining that the read-aloud voice is unavailable; In the case that the reading voice is unavailable, corresponding reading correction information is output according to the unavailable type of the reading voice, wherein the unavailable type includes mismatch of sound attributes, incomplete content of the reading voice, and the reading voice is noise.

6. The voiceprint registration method according to any one of claims 1 to 5, in, In a case where the text content corresponding to the read-aloud voice includes the read-aloud text, the text content corresponding to the read-aloud voice matches the read-aloud text.

7. The voiceprint registration method according to claim 1, further comprising: include: Acquiring the sound attribute in the registration information input by the user as the preset sound attribute; or, Before generating the guidance information, the user's voice is collected, and the sound attribute of the collected user's voice is determined as the preset sound attribute.

8. The voiceprint registration method according to claim 1, in, Determining the sound attribute of the read aloud speech includes: The voice features of the read-aloud speech are input into a preset sound attribute classification model to obtain the sound attributes of the read-aloud speech.

9. The voiceprint registration method according to claim 1, in, Determining the sound attribute of the read aloud speech includes: Inputting the speech features of the read speech into a preset neural network model to obtain a speech embedding feature vector extracted by a hidden layer of the neural network model; Calculating the distance between the speech embedding feature vector and a preset speech embedding feature vector of each sound attribute, and determining the shortest distance therebetween; When the shortest distance is not greater than a preset distance threshold, determining the sound attribute corresponding to the shortest distance as the sound attribute of the read-aloud voice; When the shortest distance is greater than a preset distance threshold, the sound attribute of the read-aloud speech is determined as an unknown attribute.

10. The voiceprint registration method according to claim 1, in, The identifying of text content corresponding to the read voice includes: The speech recognition model corresponding to the sound attribute of the read speech is used to determine the text content corresponding to the read speech.

11. The voiceprint registration method according to claim 1, in, The isolated word recognition model corresponding to the preset sound attribute is used to recognize the read-aloud voice, and the text content corresponding to the read-aloud voice is obtained.

12. The voiceprint registration method according to claim 1, in, The guidance information is a guidance voice, and the process text includes a description text corresponding to each randomly selected word.

13. The voiceprint registration method according to claim 1, in, The sound attribute is a dialect type.

14. The voiceprint registration method according to any one of claims 1 to 5, further comprising: include: For the guidance information for voiceprint verification, when the reading voice is available, if the extracted voiceprint matches the stored voiceprint corresponding to the user, the identity verification of the user is passed.

15. A voiceprint registration device, include: A reading text generation module is configured to randomly select a preset number of characters from a character library to form candidate texts, and determine a phoneme combination of the candidate texts; Detecting the occurrence frequency of the phoneme combination in a preset text library; when the occurrence frequency of the phoneme combination is lower than a preset frequency, using the candidate text as the read text; a guidance information generating module, configured to generate guidance information according to the read text and a process text for guiding the user to speak the read text with a preset sound attribute, wherein the preset sound attribute is a sound attribute in the user's registration information or a sound attribute of the user's voice collected before generating the guidance information; A reading voice acquisition module, configured to acquire the user's reading voice after the guidance information is output to the user; A reading speech analysis module, configured to determine the sound attributes of the reading speech and recognize the text content corresponding to the reading speech; an availability determination module, configured to determine that the reading voice is available when the sound attribute of the reading voice is the preset sound attribute and the text content corresponding to the reading voice matches the reading text, wherein the text content corresponding to the reading voice matches the reading text when the phoneme sequence corresponding to the text content corresponding to the reading voice includes the phoneme sequence corresponding to the reading text; The storage module is configured to store the corresponding relationship between the user and the voiceprint information extracted from the reading voice for the guidance information for voiceprint registration when the reading voice is available.

16. A voiceprint registration system, include: The voiceprint registration device according to claim 15; an output device, configured to output the guidance information generated by the voiceprint registration device; as well as The recording device is configured to record the user's spoken voice.

17. The voiceprint registration system according to claim 16, in, The output device is a sound output device.

18. A voiceprint registration device, include: Memory; as well as A processor coupled to the memory, the processor being configured to execute the voiceprint registration method according to any one of claims 1 to 14 based on instructions stored in the memory.

19. A computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, the voiceprint registration method according to any one of claims 1 to 14 is implemented.

Images