WO2018137426A1 - Method and apparatus for recognizing voice information of user - Google Patents

Abstract

Provided are a method and apparatus for recognizing voice information of a user. The method comprises: obtaining voice information of a user; extracting first language features and voice features of the voice information; searching for preset second language features corresponding to the voice features; and determining whether the voice information is legitimate according to a first comparison result of the first language features and the second language features.

Description

User voice information identification method and device Technical field

The present disclosure relates to the field of communications, and in particular to a method and apparatus for identifying user voice information.

Background technique

The identity recognition system includes face recognition, fingerprint recognition, voiceprint recognition, password recognition and password recognition.

Password identification means that the user inputs or speaks one or more words, phrases or sentences as the key for verification, but the identification technology can only be used for simple identification, and the key is easily stolen or stolen.

Voiceprint is a sound wave spectrum that carries speech information displayed by electroacoustic instruments. The generation of human language is a complex physiological and physical process between the human language center and the vocal organs. The vocal organs used by people when speaking, for example, The tongue, teeth, throat, lungs, and nasal cavity vary greatly in size and morphology, so there are differences in the voiceprint of any two people. Although each person's phonetic characteristics are not static and variability, they are affected by physical, pathological, psychological, analog, camouflage, and environmental factors, but voiceprint recognition technology is more secure than password recognition technology.

However, only the use of voiceprint recognition technology for the identification of user identity is still a security risk, because the phonetic features can still be imitated by certain techniques, such as illegally accepting the voice information of the user's speech, through the pirated The voice information mimics the user's voice, causing the user's identity information to be stolen, causing economic losses to the user.

Summary of the invention

The following is an overview of the topics detailed in this document. This Summary is not intended to limit the scope of the claims.

Embodiments of the present disclosure provide a method and apparatus for identifying user voice information.

According to an embodiment of the present disclosure, a method for identifying user voice information is provided, including: Acquiring the voice information of the user; extracting the first language feature and the voice feature of the voice information; searching for a preset second language feature corresponding to the voice feature; according to the first language feature and the second language feature The first comparison result determines whether the sound information is legal.

In an exemplary embodiment, determining whether the sound information is legal according to the first comparison result of the first language feature and the second language feature comprises: determining the first language feature according to the first comparison result a vector similarity with the second language feature; determining whether the sound information is legal according to a comparison result of the vector similarity and a preset threshold, wherein when the comparison result indicates that the vector similarity is greater than or When the preset threshold is equal to, the sound information is determined to be legal.

In an exemplary embodiment, the preset threshold is an average of vector similarities between a plurality of the second language features.

In an exemplary embodiment, the searching for the preset second language feature corresponding to the voice feature comprises: searching for user identification information corresponding to the voice feature; and acquiring the first according to the found user identifier information Two language features.

In an exemplary embodiment, the correspondence between the voice feature and the user identification information is determined by using a voice feature and user identification information input in advance as an input of a neural network model in the neural network model. Train the learning to get the corresponding relationship.

In an exemplary embodiment, the first linguistic feature and the second linguistic feature are Mel Frequency Cepstral Coefficents (MFCC); the speech feature is a linear cepstrum coefficient (Linear Prediction Cepstrum Coefficient) , referred to as LPCC).

In an exemplary embodiment, the first linguistic feature and the second voicing feature comprise a voiceprint feature; the voice feature comprising linguistic content of a user password.

According to another embodiment of the present disclosure, there is provided an apparatus for identifying user voice information, comprising: an obtaining module configured to acquire sound information of a user; and an extracting module configured to extract a first language feature and a voice of the sound information a locating module configured to find a pre-set second language feature corresponding to the voice feature; the determining module configured to determine the first comparison result according to the first language feature and the second language feature Whether the sound information is legal.

In an exemplary embodiment, the determining module is further configured to determine according to the first comparison result a vector similarity between the first language feature and the second language feature; and determining whether the sound information is legal according to a comparison result of the vector similarity and a preset threshold, wherein when the comparison result indicates When the vector similarity is greater than or equal to the preset threshold, it is determined that the sound information is legal.

In an exemplary embodiment, the lookup module is further configured to look up user identification information corresponding to the voice feature; and acquire the second language feature according to the found user identification information.

According to another embodiment of the present disclosure, a storage medium is also provided. The storage medium is configured to store program code for performing the steps of: acquiring sound information of the user; extracting first language features and voice features of the sound information; and searching for pre-set second language features corresponding to the voice features Determining whether the sound information is legal according to the first comparison result of the first language feature and the second language feature.

In an exemplary embodiment, the storage medium is further configured to store program code for performing a step of determining a vector similarity between the first language feature and the second language feature based on the first comparison result; Determining whether the sound information is legal according to the comparison result of the vector similarity and the preset threshold, wherein when the comparison result indicates that the vector similarity is greater than or equal to the preset threshold, determining that the sound information is legal .

In an exemplary embodiment, the storage medium is further configured to store program code for performing: searching for user identification information corresponding to the voice feature; acquiring the second language feature based on the found user identification information .

In an exemplary embodiment, the storage medium is further configured to store program code for performing training learning in the neural network model with pre-entered speech features and user identification information as inputs to a neural network model, The corresponding relationship is obtained.

Through the present disclosure, since the linguistic features and the voice features of the sound information are acquired and extracted, and the corresponding linguistic features are searched according to the voice features, whether the sound information is legal according to the comparison result between the linguistic features is determined, and the identification process is implemented. The linguistic features and voice features of the user's voice information are comprehensively considered. With the solution of the present disclosure, it is solved that there is currently no recognition method combining the password recognition and the voiceprint recognition technology to comprehensively consider the language feature and the voice feature of the user, and the key of the password recognition technology or the voiceprint recognition technology is relatively easy to be used by others. Stealing or stealing, and then giving the user The problem of economic loss and poor user experience effectively combines password recognition technology and voiceprint recognition technology, and improves the security of identity recognition by comprehensively considering the language features and voice features of the user's voice information. The probability that a key is stolen or stolen by others increases the user experience.

Other aspects will be apparent upon reading and understanding the drawings and detailed description.

DRAWINGS

1 is a block diagram showing the hardware structure of a computer terminal for identifying a user voice information according to an embodiment of the present disclosure;

2 is a flowchart of a method of identifying user sound information according to an embodiment of the present disclosure;

FIG. 3 is a structural block diagram of an apparatus for identifying user sound information according to an exemplary embodiment of the present disclosure; FIG.

FIG. 4 is a flowchart (1) of a method of identifying user sound information according to an exemplary embodiment of the present disclosure;

FIG. 5 is a flowchart (2) of a method of identifying user sound information according to an exemplary embodiment of the present disclosure;

FIG. 6 is a structural block diagram of an apparatus for identifying user sound information according to an embodiment of the present disclosure.

detailed description

At present, there is no recognition method combining the password recognition technology and the voiceprint recognition technology to comprehensively consider the user's language features and voice features, and the existing password recognition technology or voiceprint recognition technology key is easily stolen or stolen by others. Then, the user is economically lost, and the user experience is poor.

The present disclosure will be described in detail below with reference to the drawings in conjunction with the embodiments.

The terms "first", "second" and the like in the specification and claims of the present disclosure and the above-mentioned figures are used to distinguish similar objects, and are not necessarily used to describe a particular order or order.

In order to facilitate the understanding of the embodiments of the present disclosure, the following technical terms are involved in the embodiments of the present disclosure. Interpretation is as follows:

Vector similarity: is the similarity between linguistic feature vectors, usually expressed by the distance between vectors, such as Euclidean distance, cosine distance, and so on.

Example 1

The method embodiment provided by Example 1 of the present disclosure may be executed in a mobile terminal, a computer terminal, or the like. Taking a computer terminal as an example, FIG. 1 is a hardware structural block diagram of a computer terminal for identifying a user voice information according to an embodiment of the present disclosure. As shown in FIG. 1, computer terminal 10 may include one or more (only one shown) processor 102 (processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA) A memory 104 configured to store data, and a transmission device 106 for communication functions. It will be understood by those skilled in the art that the structure shown in FIG. 1 is merely illustrative and does not limit the structure of the above electronic device. For example, computer terminal 10 may also include more or fewer components than those shown in FIG. 1, or have a different configuration than that shown in FIG.

The memory 104 can be configured as a software program and a module for storing application software, such as program instructions/modules corresponding to the method for identifying user sound information in the embodiment of the present disclosure, and the processor 102 runs the software program and module stored in the memory 104. Thereby performing various functional applications and data processing, that is, implementing the above method. Memory 104 may include high speed random access memory, and may also include non-volatile memory such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory. In some examples, memory 104 may further include memory remotely located relative to processor 102, which may be coupled to computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

Transmission device 106 is configured to receive or transmit data via a network. The network specific examples described above may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC) that can be connected to other network devices through a base station to communicate with the Internet. In one example, the transmission device 106 can be a Radio Frequency (RF) module configured to communicate with the Internet wirelessly.

In this example, a method for identifying user voice information running in the computer terminal shown in FIG. 1 is provided. FIG. 2 is a flowchart of a method for identifying user voice information according to an embodiment of the present disclosure, as shown in FIG. The process includes the following steps:

Step S202, acquiring sound information of the user;

Step S204, extracting first language features and voice features of the sound information;

Step S206, searching for a preset second language feature corresponding to the voice feature;

Step S208: Determine whether the sound information is legal according to the first comparison result of the first language feature and the second language feature.

In this step, the sound information includes a language feature and a voice feature, wherein the language feature is a specific content of the user password, and the voice feature is a voiceprint feature extracted from the sound information. For example, when the user issues the password "turn on the light", the word "turning on the light" is the linguistic feature of the user's voice information, and the feature of the sound emitted by the user through the vocal organ is the voice feature.

Through the above steps, since the linguistic features and the voice features of the sound information are acquired and extracted, and the corresponding linguistic features are searched according to the voice features, whether the sound information is legal according to the comparison result between the linguistic features is determined, and the identification process is implemented. The linguistic features and voice features of the user's voice information are comprehensively considered. With the solution of the present disclosure, it is solved that there is currently no recognition method combining the password recognition and the voiceprint recognition technology to comprehensively consider the language feature and the voice feature of the user, and the key of the password recognition technology or the voiceprint recognition technology is relatively easy to be used by others. Stealing or stealing, and then giving users economic losses and poor user experience, effectively combining password recognition technology with voiceprint recognition technology, and improving the language and voice characteristics of the user's voice information. The security of identity greatly reduces the probability that the key will be stolen or stolen by others, which improves the user experience.

In an exemplary embodiment, the execution body of the above steps may be a data processing device such as a single chip microcomputer, but is not limited thereto.

In an exemplary embodiment, step S208 may be performed by determining a vector similarity between the first language feature and the second language feature according to the first comparison result; comparing the vector similarity with the preset threshold As a result, it is determined whether the sound information is legal, wherein when the comparison result indicates that the vector similarity is greater than or equal to the preset threshold, it is determined that the sound information is legal.

In this example, the second language feature may be a plurality of language features, and when the first language feature is compared with the second language feature, the vector similarity between the first language feature and each second language feature may be sequentially determined. And then take the average of the vector similarity, but is not limited to this. Because this example uses the method of vector similarity and threshold comparison to judge whether the sound information is legal, it realizes the fast and accurate judgment of the user's identity, and improves the accuracy of identifying the user's identity according to the language feature.

In an exemplary embodiment, the predetermined threshold is an average of vector similarities between the plurality of second language features. In this example, in the process of setting the threshold, the language feature of the user may be repeatedly collected, and then the average of the vector similarity between the collected language features is obtained, and the average objectively and accurately reflects the user. The linguistic features of the sound information prevent the problem of inaccurate acquisition results due to interference with the language features of a certain acquisition.

In an exemplary embodiment, step S206 may be performed by searching for user identification information corresponding to the voice feature, and acquiring the second language feature according to the found user identification information.

In this example, the user identification information may be composed of an integer string or a string of names, typically the ID of the user, which is assigned by the system and is unique.

In an exemplary embodiment, the correspondence between the voice feature and the user identification information is determined by using the voice feature and the user identification information input in advance as input of a neural network model, and performing training learning in the neural network model. Get the corresponding relationship.

In this example, the neural network model may be a deep neural network recognition model, such as a Convolutional Neural Network (CNN). During the initialization process, the deep neural network model is pre-trained. In the pre-training, there must be a large training set. The content of the training set is the pre-entered speech feature and user identification information. The pre-training process is a standard deep neural network training process, and the content of the training set is input into the depth one by one. In the neural network, the parameters of the neural network are optimized by the Error Back Propagation (BP) algorithm according to the comparison of the output and the user identification information, and the loop is performed until the accuracy of the model reaches the requirement. By adopting the neural network model and the training learning method, the correspondence between the voice feature and the user identification information can be effectively obtained, and the voice feature corresponding to the user's voice information can be quickly and accurately searched for the corresponding user identification information.

In an exemplary embodiment, the first linguistic feature and the second linguistic feature are MFCC; the speech feature is LPCC.

In this example, the speech feature can use LPCC as the feature parameter, and the LPCC parameter has the advantage of high computational efficiency, and the excitation information in the speech generation process is completely removed, which effectively reflects the channel response, and requires less LPCC. More than a dozen LPCCs can describe the formant characteristics of speech signals.

In an exemplary embodiment, the first language feature and the second phone feature include a voiceprint feature; the voice feature includes linguistic content of a user password.

3 is a structural block diagram of an apparatus for identifying user sound information according to an exemplary embodiment of the present disclosure. As shown in FIG. 3, the apparatus includes: a sound entry module 32 configured to collect sound information; and a language feature recognition module 34 configured to The language feature is extracted from the voice information; the voice feature recognition module 36 is configured to extract the voice feature in the sound information; the feature comparison module 38 is configured to compare the language feature and the vector similarity; and the feature storage module 310 is configured to store the preset The linguistic feature, the user identification information corresponding to the linguistic feature, the preset threshold, and the correspondence between the voice feature and the user identification information; the recognition result output module 312 is configured to output the indication information that determines whether the voice information is legal.

FIG. 4 is a flowchart (1) of a method for identifying user voice information according to an exemplary embodiment of the present disclosure. As shown in FIG. 4, the flow includes:

Step S402, the sound input module 32 collects the sound information of the password of the user;

Step S404, the language feature recognition module 34 performs Fourier transform on the sound information to obtain the spectrum of the sound information, and takes the logarithm of the spectrum and then performs inverse Fourier transform to obtain the MFCC;

Step S406, the sound entry module 32 and the language feature recognition module 34 repeat the above-mentioned acquisition and acquisition operations until n MFCCs are acquired, where n≥2;

Step S408, the feature comparison module 38 compares n MFCCs in pairs, obtains multiple vector similarities, and averages the plurality of vector similarities to obtain a preset threshold.

Step S410, the speech feature recognition module 36 extracts the LPCC in the sound information; inputs the LPCC and the user ID into the pre-trained deep neural network model, and performs training learning on the deep neural network model;

Step S412, the sound input module 32 and the voice feature recognition module 36 repeat the above-mentioned acquisition and training learning operations until the test effect of the deep neural network model reaches the specified requirement;

In step S414, the storage module 310 stores n MFCCs, IDs of users corresponding to the MFCCs, preset thresholds, and correspondences between the LPCCs and the IDs of the users.

This example is an operation in which the device configures relevant parameters when the user first uses the identification device of the user's voice information. In this example, the sound entry module 32 can perform sound information collection through the microphone.

FIG. 5 is a flowchart (2) of a method for identifying user voice information according to an exemplary embodiment of the present disclosure. As shown in FIG. 5, the flow includes:

Step S502, the sound input module 32 collects the sound information of the password of the user;

Step S504, the linguistic feature recognition module 34 performs Fourier transform on the sound information to obtain the spectrum of the sound information, and takes the logarithm of the spectrum and then performs inverse Fourier transform to obtain the MFCC;

Step S506, the speech feature recognition module 36 extracts the LPCC in the sound information; inputs the LPCC into the pre-trained deep neural network model to obtain the ID of the user;

Step S508, the feature comparison module 38 searches the storage module 310 for the MFCC corresponding to the obtained user ID, and compares the found MFCC with the MFCC extracted from the voice information of the password password to obtain a vector similarity; The vector similarity is compared with the preset threshold. When the vector similarity is greater than or equal to the preset threshold, the voice information of the password password is legal, the user identification is successful; when the vector similarity is less than the preset threshold, the voice information of the password password Is illegal, user identification fails;

In step S510, the recognition result outputting module 312 outputs the recognition result of the successful or failed user identification determined by the feature comparison module 38.

The above method steps of the present example can be applied to the following scenarios, but are not limited thereto.

When the user returns home, an "open door" command is issued at the door of the door; the identification device of the user's voice information prompts for a password, and the user speaks the preset password password "I am the homeowner"; the identification device enters the password and the voice according to the password The feature determines a preset language feature; the identifying device compares the language feature extracted by the password and the password with the preset language feature to obtain a vector similarity; compares the vector similarity with the preset threshold, and when the vector similarity is greater than or equal to the preset When the threshold is reached, the door is automatically opened; When the vector similarity is less than the preset threshold, the user is prompted to re-enter the password.

Through the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware, but in many cases, the former is The usual implementation. Based on this understanding, the solution of the present disclosure may be embodied in the form of a software product stored in a storage medium (such as a ROM/RAM, a magnetic disk, an optical disk), and includes a plurality of instructions for making one The terminal device (which may be a cell phone, computer, server, or network device, etc.) performs the methods described in various embodiments of the present disclosure.

Example 2

Also provided in this example is an identification device for user sound information, which is used to implement the previous examples and exemplary embodiments, which are not described again. As used below, the term "module" may implement a combination of software and/or hardware of a predetermined function. Although the devices described in the following embodiments are typically implemented in software, hardware, or a combination of software and hardware, is also possible and contemplated.

6 is a structural block diagram of an apparatus for identifying user voice information according to an embodiment of the present disclosure. As shown in FIG. 6, the apparatus includes an obtaining module 62 configured to acquire sound information of a user, and an extracting module 64 configured to extract sound information. a first language feature and a voice feature; the search module 66 is configured to search for a preset second language feature corresponding to the voice feature; the determining module 68 is configured to determine according to the first comparison result of the first language feature and the second language feature Whether the sound information is legal.

In an exemplary embodiment, the determining module 68 is further configured to determine a vector similarity between the first language feature and the second language feature according to the first comparison result; and compare the vector similarity with the preset threshold. Determining whether the sound information is legal, wherein when the comparison result indicates that the vector similarity is greater than or equal to the preset threshold, it is determined that the sound information is legal.

In an exemplary embodiment, the lookup module 66 is further configured to search for user identification information corresponding to the voice feature; and acquire the second language feature according to the found user identification information.

This example differs from Example 1 in the division of the module. In this example, the acquisition module 62 is similar to the sound entry module 32 in Example 1, but with some new functional features added; Block 64 is similar to language feature recognition module 34 and speech feature recognition module 36 in Example 1, but with some new functional features added; feature comparison module in Example 1 is implemented by lookup module 66 and determination module 68 in this example. 38 features.

The above modules may be implemented by software or hardware. For the latter, the foregoing may be implemented by, but not limited to, the above modules are all located in the same processor; or, the above modules are respectively located in different combinations. In the processor.

Example 3

Embodiments of the present disclosure also provide a storage medium. In the present example, the above storage medium may be configured to store program code for performing the following steps: S11, acquiring sound information of the user; S12, extracting first language features and voice features of the sound information; S13, finding and voice features Corresponding pre-set second language features; S14, determining whether the sound information is legal according to the first comparison result of the first language feature and the second language feature.

In this example, the storage medium may be further configured to store program code for performing the following steps: S21, determining a vector similarity between the first language feature and the second language feature according to the first comparison result; S22, according to the vector The result of the comparison between the similarity and the preset threshold determines whether the sound information is legal. When the comparison result indicates that the vector similarity is greater than or equal to the preset threshold, the sound information is determined to be legal.

In this example, the storage medium may be further configured to store program code for performing the following steps: S31, searching for user identification information corresponding to the voice feature; S32, acquiring the second language feature based on the found user identification information.

In this example, the foregoing storage medium may include, but is not limited to, a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk. Or a variety of media such as optical discs that can store program code.

For specific embodiments in this example, reference may be made to the examples described in the previous examples and exemplary embodiments, and details are not described herein again.

One of ordinary skill in the art will appreciate all or some of the steps in the methods disclosed above. The functional modules/units in the system, device, and device can be implemented as software, firmware, hardware, and suitable combinations thereof. In a hardware implementation, the division between functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be composed of several physical The components work together. Some or all of the components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer readable medium, which may include computer storage media (or non-transitory media) and communication media (or transitory media). As is well known to those of ordinary skill in the art, the term computer storage medium includes volatile and nonvolatile, implemented in any method or technology for storing information, such as computer readable instructions, data structures, program modules or other data. Sex, removable and non-removable media. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridge, magnetic tape, magnetic disk storage or other magnetic storage device, or may Any other medium used to store the desired information and that can be accessed by the computer. Moreover, it is well known to those skilled in the art that communication media typically includes computer readable instructions, data structures, program modules, or other data in a modulated data signal, such as a carrier wave or other transport mechanism, and can include any information delivery media. .

The above description is only exemplary embodiments of the present disclosure, and is not intended to limit the disclosure, and various changes and modifications may be made to the present disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and scope of the present disclosure are intended to be included within the scope of the present disclosure.

Industrial applicability

Through the present disclosure, since the linguistic features and the voice features of the sound information are acquired and extracted, and the corresponding linguistic features are searched according to the voice features, whether the sound information is legal according to the comparison result between the linguistic features is determined, and the identification process is implemented. The linguistic features and voice features of the user's voice information are comprehensively considered. With the solution of the present disclosure, it is solved that there is currently no recognition method combining the password recognition and the voiceprint recognition technology to comprehensively consider the language feature and the voice feature of the user, and the key of the password recognition technology or the voiceprint recognition technology is relatively easy to be used by others. Stealing or stealing, and then giving the user The problem of economic loss and poor user experience effectively combines password recognition technology and voiceprint recognition technology, and improves the security of identity recognition by comprehensively considering the language features and voice features of the user's voice information. The probability that a key is stolen or stolen by others increases the user experience. The present disclosure therefore has industrial applicability.

Claims (11)

A method for identifying user voice information, comprising: Obtaining the user's voice information; Extracting first language features and voice features of the sound information; Finding a preset second language feature corresponding to the voice feature; Determining whether the sound information is legal according to the first comparison result of the first language feature and the second language feature. The method of claim 1, wherein determining whether the sound information is legal according to the first comparison result of the first language feature and the second language feature comprises: Determining a vector similarity between the first language feature and the second language feature according to the first comparison result; Determining whether the sound information is legal according to the comparison result of the vector similarity and the preset threshold, wherein when the comparison result indicates that the vector similarity is greater than or equal to the preset threshold, determining that the sound information is legal . The method of claim 2, wherein the predetermined threshold is an average of vector similarities between a plurality of the second language features. The method of claim 1, wherein the finding a pre-set second language feature corresponding to the voice feature comprises: Finding user identification information corresponding to the voice feature; And acquiring the second language feature according to the found user identification information. The method according to claim 4, wherein the correspondence between the voice feature and the user identification information is determined by: The voice feature and the user identification information input in advance are used as input of a neural network model, and training learning is performed in the neural network model to obtain the corresponding relationship. The method according to any one of claims 1 to 5, wherein The first language feature and the second phone feature are a Mel spectrum reciprocal MFCC; The speech feature is a linear cepstral coefficient LPCC. The method according to any one of claims 1 to 5, wherein The first language feature and the second phone feature include a voiceprint feature; The speech feature includes linguistic content of the user's password. A device for identifying user voice information, comprising: An obtaining module (62) configured to obtain sound information of the user; An extracting module (64) configured to extract a first language feature and a voice feature of the sound information; a lookup module (66) configured to look up a pre-set second language feature corresponding to the voice feature; The determining module (68) is configured to determine whether the sound information is legal according to the first comparison result of the first language feature and the second language feature. The apparatus of claim 8, wherein the determining module (68) is further configured to determine a vector similarity between the first language feature and the second language feature based on the first comparison result; Determining whether the sound information is legal according to the comparison result of the vector similarity and the preset threshold, wherein when the comparison result indicates that the vector similarity is greater than or equal to the preset threshold, determining that the sound information is legal . The apparatus of claim 8, wherein the lookup module (66) is further configured to look up user identification information corresponding to the voice feature; and to acquire the second language feature based on the found user identification information . A computer readable storage medium storing computer executable instructions that, when executed by a processor, implement the method of any of claims 1-7.

Images