CN114387974A - Case-crossing method, system, device and storage medium based on voiceprint recognition - Google Patents

Case-crossing method, system, device and storage medium based on voiceprint recognition Download PDF

Info

Publication number
CN114387974A
CN114387974A CN202111516627.9A CN202111516627A CN114387974A CN 114387974 A CN114387974 A CN 114387974A CN 202111516627 A CN202111516627 A CN 202111516627A CN 114387974 A CN114387974 A CN 114387974A
Authority
CN
China
Prior art keywords
voice
sample
voiceprint
voices
similar
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202111516627.9A
Other languages
Chinese (zh)
Inventor
郑骁敏
肖龙源
李稀敏
邓仁超
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xiamen Kuaishangtong Technology Co Ltd
Original Assignee
Xiamen Kuaishangtong Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xiamen Kuaishangtong Technology Co Ltd filed Critical Xiamen Kuaishangtong Technology Co Ltd
Priority to CN202111516627.9A priority Critical patent/CN114387974A/en
Publication of CN114387974A publication Critical patent/CN114387974A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L17/00—Speaker identification or verification techniques
    • G10L17/02—Preprocessing operations, e.g. segment selection; Pattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal components; Feature selection or extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/242—Query formulation
    • G06F16/2433—Query languages
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/242—Query formulation
    • G06F16/2433—Query languages
    • G06F16/244—Grouping and aggregation
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
    • G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M3/00—Automatic or semi-automatic exchanges
    • H04M3/22—Arrangements for supervision, monitoring or testing
    • H04M3/2281—Call monitoring, e.g. for law enforcement purposes; Call tracing; Detection or prevention of malicious calls

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Human Computer Interaction (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computer Security & Cryptography (AREA)
  • Technology Law (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention discloses a method, a system, a device and a storage medium for case string based on voiceprint recognition, wherein the system comprises the following steps: the voice print library is used for storing voice print vectors of sample voice; the vector module is used for acquiring a voiceprint vector of the sample voice and the retrieval voice; and the analysis module is used for searching sample voice with similar voiceprint vectors to the searched voice in the voiceprint library to obtain similar sample voice. Aiming at the telecommunication case, the invention can solve the problem that the objects cannot be associated due to the lack of label data, increase the association between the voiceprint vectors of case voice data and improve the convenience of case string.

Description

Case-crossing method, system, device and storage medium based on voiceprint recognition
Technical Field
The invention relates to the technical field of voiceprint recognition calculation, in particular to a method, a system, a device and a storage medium for case string based on voiceprint recognition.
Background
For cases completed by the same person with different identity information, in the existing method, more cases are struggled through information tags, namely, a large number of tags are marked on case completers or related cases, and the cases completed by the same person are associated through information tags of case performing regions, case contacts, case means and the like, so that the cases completed by the same person are found.
Although the relevance is checked in a labeling mode and the relevance may be directed to the same object, how to establish the relevance of different information labels is a technical problem in the case checking process. Secondly, most of the current cases are carried out through telecommunication, and the cases are difficult to label through a human face recognition way. Accordingly, there is a need for a method that facilitates determining associations between different cases for string analysis.
Disclosure of Invention
Aiming at solving the problems, the invention provides a method, a system, a device and a storage medium for case string based on voiceprint recognition, which can solve the problem that objects cannot be associated due to lack of label data aiming at a telecommunication case, increase the association among voiceprint vectors of case voice data and improve the convenience of case string.
In order to achieve the purpose, the invention adopts the technical scheme that:
a system for a string case based on voiceprint recognition, comprising: the voice print library is used for storing voice print vectors of sample voice; the vector module is used for acquiring the voiceprint vectors of the sample voice and the retrieval voice; and the analysis module is used for searching the sample voice which has similar voiceprint vectors with the searched voice in the voiceprint library to obtain similar sample voice.
Preferably, the system further comprises a tag module and a tag screening module, wherein the tag module is used for tagging information tags on the sample voice or the retrieval voice; the label screening module is used for retrieving the sample voice with the information label correlation degree with the retrieved voice being greater than a first threshold value in the similar sample voice to obtain correlated sample voice.
Preferably, the system further includes a review module, configured to compare the voiceprint vectors of the sample voices in the similar sample voices or the associated sample voices one by one, and reject the sample voice with a poor comparison result.
Preferably, the system further comprises a voiceprint clustering module, configured to cluster the sample voices in the voiceprint library according to the similarity of the voiceprint vectors to obtain multiple voiceprint categories; and the analysis module determines the voiceprint category to which the retrieval voice belongs according to the voiceprint vector of the retrieval voice, and outputs the sample voice in the voiceprint category as the similar sample voice.
Preferably, the analysis module includes a generalization retrieval module, configured to retrieve the sample speech in the voiceprint library, where a similarity between the voiceprint vector of the retrieved speech and the sample speech is greater than a second threshold, and obtain a generalized sample speech as the similar sample speech.
Preferably, the analysis module includes a cyclic retrieval module, configured to retrieve the sample voices in the voiceprint library, of which similarity to the voiceprint vector of the retrieved voice is greater than a first threshold, select the first N sample voices according to the similarity ranking, and obtain first similar sample voices; then, the sample voices in the first similar sample voice are retrieved one by one, the sample voices, the similarity of which with the voiceprint vector of the sample voice of the first sample voice is larger than a first threshold value, in the voiceprint library are retrieved, the sample voices are sorted according to the similarity, and the first N sample voices are selected to obtain a second similar sample voice; then, the sample voices in the second similar sample voice are retrieved one by one, the sample voices, the similarity of which with the voiceprint vector of the sample voice of the second sample voice is larger than a first threshold value, in the voiceprint library are retrieved, the sample voices are sorted according to the similarity, and the first N sample voices are selected to obtain a third similar sample voice; circulating for M times; taking the first similar voice, the second similar voice, … … and the Mth similar voice as the similar voice; n, M are positive integers.
Preferably, the analysis module includes a cyclic retrieval module, configured to retrieve the sample voices in the voiceprint library, of which similarity to the voiceprint vector of the retrieved voice is greater than a first threshold, sort the sample voices according to the similarity, select the first N sample voices, and add the sample voices to the sample voices; then, the sample voices in the similar sample voices are retrieved one by one, the sample voices, the similarity of which with the voiceprint vector of the sample voice of the similar sample voice is larger than a first threshold value, in the voiceprint library are retrieved, the sample voices are sorted according to the similarity, and the first N sample voices are selected as retrieval results; if the sample voice in the retrieval result is not in the similar sample voice, the retrieval result is not added into the similar sample voice, otherwise, the retrieval result is added into the similar sample voice; and repeating the steps, wherein N is respectively a positive integer.
Based on the same inventive concept, the invention also provides a case crossing device based on voiceprint recognition, which comprises: an input terminal for inputting a sample voice or a retrieval voice; and the searching terminal is loaded with the system and is used for obtaining the similar sample voice or the related sample voice.
Based on the same inventive concept, the invention also provides a case-crossing method based on voiceprint recognition, which comprises the following steps: acquiring and storing a voiceprint vector of the sample voice; acquiring a voiceprint vector of retrieval voice; and retrieving the sample voice with the similar voiceprint vector with the retrieved voice in the voiceprint library to obtain the similar sample voice.
Based on the same inventive concept, the present invention also provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the above-mentioned method.
The invention has the beneficial effects that:
1. creating a voiceprint library, and increasing a voiceprint vector as a basis for string case retrieval, so that the convenience of string cases is improved;
2. the retrieval result based on voiceprint recognition is screened by using the information tag, so that the utilization rate of tag information and the accuracy of the result output by the system are improved;
3. comparing and rechecking the retrieval result based on voiceprint recognition, and rejecting sample voices with poor comparison results;
4. and a multi-round cyclic retrieval method is adopted, so that the accuracy of the retrieval result based on voiceprint recognition is improved.
Drawings
FIG. 1 is a flowchart illustrating operation of a loop search module according to an embodiment;
FIG. 2 is a flow chart of the method according to the second embodiment.
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects to be solved by the present invention clearer and more obvious, the present invention is further described in detail with reference to specific embodiments below. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Example one
This embodiment provides a string case system based on voiceprint recognition, includes:
and the voiceprint library is created based on the sample voice and is used for storing the voiceprint vector of the sample voice.
The sample voice described in this embodiment is a voice used for registering a voiceprint vector in a voiceprint library. The source of the sample speech is usually from case-collected speech data, or from conventionally registered speech data.
The retrieval voice described in this embodiment refers to a voice for retrieving a source of a misstatement. After the retrieval voice is used, the retrieval voice is registered in the voiceprint library as sample voice, so that the voiceprint library is expanded, and the coverage of the voiceprint library on cases is perfected.
And the vector module is used for acquiring the voiceprint vectors of the sample voice and the retrieval voice so as to retrieve in the voiceprint library by using the voiceprint vectors of the retrieval voice.
The method comprises the following steps of analyzing a string of cases, namely finding out the phenomenon that single-person multi-voice counterfeits multi-person multi-voice in voice data based on the angle of the case, and finding out a set that a plurality of pieces of voice data point to the same person according to the result of the step. The system of this embodiment can perform the string pattern analysis by using the three methods as described in the first to third embodiments, respectively.
In this embodiment, a voiceprint clustering method is used to perform the string pattern analysis.
The system of this embodiment further includes a voiceprint clustering module configured to cluster the sample voices in the voiceprint library according to the similarity of the voiceprint vectors to obtain a plurality of voiceprint categories.
The meaning of the voiceprint clustering is that sample voices with the similarity of voiceprint vectors in a voiceprint library reaching a set threshold value are aggregated together to form different voiceprint categories. The voiceprint clustering method is actually a voiceprint self-checking operation, and has the advantages that the voice does not need to be retrieved and can be executed regularly. When the voiceprint string pattern analysis is needed each time, the result of the retrieved voice and the voiceprint clustering is only needed to be compared.
And the analysis module is used for searching sample voice with similar voiceprint vectors to the searched voice in the voiceprint library to obtain similar sample voice. In the present embodiment, the voiceprint class to which the search speech belongs is determined according to the voiceprint vector of the search speech, and the sample speech in the voiceprint class is output as the similar sample speech.
And the rechecking module is used for comparing the voiceprint vectors of the sample voices in the similar sample voices one by one and rejecting the sample voices with poorer comparison results.
When the number of sample voices in the voiceprint library is large, the number of search results output by the analysis module may be large, so that a check is performed on the search results. The verification mode is that every two voiceprint vectors of sample voices in the retrieval result are compared for n x (n-1)/2 times, and the results with lower scores after comparison are removed.
And the rechecking result and the retrieval voice output by the rechecking module are used as the case string result.
Example two
This embodiment provides a string case system based on voiceprint recognition, includes:
the voiceprint library as described in example one.
The vector module of embodiment one.
In this embodiment, a single generalized search method is used to perform the pattern analysis.
The analysis module comprises a generalization retrieval module which is used for retrieving the sample voice with the similarity of the voiceprint vector of the retrieved voice larger than the threshold value in the voiceprint library to obtain the generalized sample voice as the similar sample voice.
Although the method is simple to implement, the number of sample voices in the voiceprint library is large, voiceprint vectors larger than a threshold value and obtained by a generalized retrieval module are many, and many similar sample voices exist correspondingly, so that a retrieval result obtained by the method is inaccurate, the similar sample voices need to be further screened by combining an information tag, and the requirement of improving the accuracy of the analysis of the string plan is met.
The system described in this embodiment also includes a tag module.
The label module is used for marking information labels on the sample voice or the retrieval voice. The label information includes region, contact person, and manipulation. Several cases can be associated by information tags and are owned by the same person.
The system described in this embodiment further includes a tag screening module. In this embodiment, relevance checking may be performed on similar sample voices through the information tag. The label screening module is used for searching sample voices, with the information label correlation degree of the searched voices larger than a threshold value, in the similar sample voices to obtain correlated sample voices. The label screening module can also perform cluster analysis on similar sample voices through the information labels.
The system described in this embodiment performs case analysis in combination with the information tag, and improves the utilization rate of tag information.
The tag module and the tag screening module described in this embodiment are not limited to be used in this embodiment, and may also be used in the first or third embodiment to improve the accuracy of the pattern analysis.
And the rechecking module is used for comparing the voiceprint vectors of the sample voices in the associated sample voices one by one and rejecting the sample voices with poorer comparison results.
And the rechecking result and the retrieval voice output by the rechecking module are used as the case string result.
EXAMPLE III
This embodiment provides a string case system based on voiceprint recognition, includes:
the voiceprint library as described in example one.
The vector module of embodiment one.
In the present embodiment, a multi-round loop search method is employed.
The analysis module comprises a cyclic retrieval module which is used for retrieving sample voices, the similarity between the sample voices and the voiceprint vector of the retrieved voice is larger than a threshold value, sorting the sample voices according to the similarity, and selecting the first N sample voices to obtain first similar sample voices.
The loop retrieval module needs to determine the number of the retrieval results of each round, for example, only the first 5, but not all, sample voices larger than the threshold are obtained for each round of retrieval and are used as the similar sample voices of the first round.
After each retrieval, the 5 sample voices are subjected to the retrieval of the second round, and so on to obtain similar sample voices of a plurality of rounds.
Conceivably, it is impossible to search all the time. The stop condition has two ways, one is to define a round, for example, to perform 3 rounds. And the other is that when the search result of a certain sample voice does not appear in the previous result set, the next round of search on the search result of the sample voice is stopped.
And taking the set of similar sample voices of all rounds as the similar sample voices output by the loop retrieval module.
Referring to fig. 1, assume that there are a-Z sample voices in the voiceprint library and that they are all different identity information. A is search speech.
The first round, make 1 in the voiceprint library with the voiceprint vector of A: and N is added. The filtering score is larger than the threshold value and is not the matching result of the filtering score, namely B and D;
and performing 1 for the sample voice B and the sample voice D respectively in the second round: and N is added. The filtering score is larger than the threshold value and is not the matching result of the filtering score, namely C, P, E and F;
and so on … …;
in round N, if none of the matching results with scores greater than the threshold are in the set of previous matching results, stop 1: and (N) operating. That is, none of X, Y, Z appears in a, B, C, D, E, F, P before, so no 1: and N is added.
And the rechecking module is used for comparing the voiceprint vectors of the sample voices in the similar sample voices one by one and rejecting the sample voices with poorer comparison results.
And the rechecking result and the retrieval voice output by the rechecking module are used as the case string result.
The voiceprint clustering method and the single generalization retrieval method can obtain results in one step, the multi-round cyclic retrieval method can obtain results only by carrying out multiple operations, and the three methods have advantages and disadvantages respectively, and the complexity is from low to high.
Aiming at the characteristics of a large number of existing telecommunication cases, a small number of marked clues and images and a large number of voices, the system provided by the invention extracts voiceprint characteristics by effectively utilizing the voices, and can realize the string case of different telecommunication cases through a voiceprint clustering method, a single generalized retrieval method or a multi-round circulating retrieval method. Meanwhile, cases which are difficult to be associated through the information tags can be associated through the voiceprint mode.
Example four
Referring to fig. 2, the present embodiment provides a method for pattern matching based on voiceprint recognition, including the following steps:
s1, obtaining and storing a voiceprint vector of a sample voice. The voiceprint vectors of the sample speech are registered in the voiceprint library.
Sample voices in the voiceprint library are tagged with information.
And S2, acquiring a voiceprint vector of the retrieval voice. And searching the searched voice in the voiceprint library as a string case source.
The information tag is printed on the retrieval voice, which can be used as a basis for further screening, so that the accuracy of the string case analysis is improved.
And S3, retrieving sample voices with similar voiceprint vectors to the retrieved voices in the voiceprint library to obtain similar sample voices, and completing retrieval by adopting a voiceprint clustering method, a single generalized retrieval method and a multi-cycle retrieval method.
In order to improve the accuracy of the string case analysis, the method for completing the search is not limited to one of the three types, and may be a combination of two or three types.
In order to improve the accuracy of the string case analysis, after the retrieval result is obtained by using the method, relevance checking can be carried out according to the retrieval voice and the information label of the retrieval result, and a more accurate result is screened out.
And S4, comparing the voiceprint vectors of the searched voices one by one, eliminating sample voices with poor comparison results, and taking the output rechecking results and the searched voices as case string results.
The present embodiment also provides a computer-readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps of the method described in the present embodiment.
EXAMPLE five
This embodiment provides a string case device based on voiceprint recognition, includes:
an input end for inputting a sample voice or retrieving a voice.
And the retrieval end is loaded with any one of the systems described in the first to third embodiments.
When the input end has voice input, the retrieval end firstly extracts and obtains the voiceprint vector of the input voice.
Secondly, if the input voice is sample voice, the retrieval end registers the input voice in a voiceprint library as the retrieved content in the string plan analysis.
If the input voice is retrieval voice, the retrieval end uses the input voice as a string case source to carry out the retrieval of the voiceprint vector similarity in the voiceprint library so as to obtain similar sample voice.
Similar sample voices can be combined with printed information labels for further information relevance screening, and the certainty that the obtained relevant sample voices point to the same object is higher, namely the accuracy of string case analysis is improved.
And thirdly, the retrieval end compares the similar sample voice or the associated sample voice for rechecking, eliminates the sample voice with poor comparison result, further reduces the range of the case-string result and improves the accuracy.
And finally, the retrieval end outputs the case string result.
While the above description shows and describes the preferred embodiments of the present invention, it is to be understood that the invention is not limited to the forms disclosed herein, but is not to be construed as excluding other embodiments and is capable of use in various other combinations, modifications, and environments and is capable of changes within the scope of the inventive concept as expressed herein, commensurate with the above teachings, or the skill or knowledge of the relevant art. And that modifications and variations may be effected by those skilled in the art without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (10)

1. A system for a string case based on voiceprint recognition, comprising:
the voice print library is used for storing voice print vectors of sample voice;
the vector module is used for acquiring the voiceprint vectors of the sample voice and the retrieval voice;
and the analysis module is used for searching the sample voice which has similar voiceprint vectors with the searched voice in the voiceprint library to obtain similar sample voice.
2. The system of claim 1, further comprising a tag module and a tag filtering module, wherein the tag module is configured to tag the sample voice or the search voice with an information tag;
the label screening module is used for retrieving the sample voice with the information label correlation degree with the retrieved voice being greater than a first threshold value in the similar sample voice to obtain correlated sample voice.
3. The system according to claim 1 or 2, further comprising a review module for comparing the voiceprint vectors of the sample voices in the similar sample voices or the associated sample voices one by one, and eliminating the sample voices with poor comparison results.
4. The system of claim 1, further comprising a voiceprint clustering module for clustering the sample speech in the voiceprint library according to the similarity of the voiceprint vectors to obtain a plurality of voiceprint categories;
and the analysis module determines the voiceprint category to which the retrieval voice belongs according to the voiceprint vector of the retrieval voice, and outputs the sample voice in the voiceprint category as the similar sample voice.
5. The system according to claim 2, wherein the analysis module comprises a generalized search module for searching the sample voices in the voiceprint library, which have a similarity greater than a second threshold with the voiceprint vector of the searched voice, to obtain generalized sample voices as the similar sample voices.
6. The system according to claim 1, wherein the analysis module comprises a cyclic search module configured to search the sample voices in the voiceprint library, the similarity of which to the voiceprint vector of the searched voice is greater than a second threshold, and select the first N sample voices according to the similarity ranking to obtain a first similar sample voice; then, the sample voices in the first similar sample voice are retrieved one by one, the sample voices, the similarity of which with the voiceprint vector of the sample voice of the first sample voice is larger than the second threshold value, in the voiceprint library are retrieved, the sample voices are sorted according to the similarity, the first N sample voices are selected, and a second similar sample voice is obtained; then, the sample voices in the second similar sample voice are retrieved one by one, the sample voices, the similarity of which with the voiceprint vector of the sample voice of the second sample voice is larger than the second threshold value, in the voiceprint library are retrieved, the sample voices are sorted according to the similarity, the first N sample voices are selected, and a third similar sample voice is obtained; circulating for M times; taking the first similar voice, the second similar voice, … … and the Mth similar voice as the similar voice; n, M are positive integers.
7. The system according to claim 1, wherein the analysis module comprises a cyclic search module for searching the sample voices in the voiceprint library with similarity to the voiceprint vector of the searched voice greater than a second threshold, and selecting the first N sample voices according to the similarity ranking, and adding the selected sample voices to the similar sample voices; then, the sample voices in the similar sample voices are retrieved one by one, the sample voices, the similarity of which with the voiceprint vector of the sample voice of the similar sample voice is larger than the second threshold value, in the voiceprint library are retrieved, the sample voices are sorted according to the similarity, and the first N sample voices are selected as retrieval results; if the sample voice in the retrieval result is not in the similar sample voice, the retrieval result is not added into the similar sample voice, otherwise, the retrieval result is added into the similar sample voice; and repeating the steps, wherein N is respectively a positive integer.
8. A device of strikeing a case based on voiceprint recognition, comprising:
an input terminal for inputting a sample voice or a retrieval voice;
a retrieval end loaded with the system as claimed in any one of claims 1 to 7 for obtaining the similar sample speech or the associated sample speech.
9. A method for case string based on voiceprint recognition is characterized by comprising the following steps:
acquiring and storing a voiceprint vector of the sample voice;
acquiring a voiceprint vector of retrieval voice;
and retrieving the sample voice with the similar voiceprint vector with the retrieved voice in the voiceprint library to obtain the similar sample voice.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method as claimed in claim 9.
CN202111516627.9A 2021-12-13 2021-12-13 Case-crossing method, system, device and storage medium based on voiceprint recognition Pending CN114387974A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111516627.9A CN114387974A (en) 2021-12-13 2021-12-13 Case-crossing method, system, device and storage medium based on voiceprint recognition

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111516627.9A CN114387974A (en) 2021-12-13 2021-12-13 Case-crossing method, system, device and storage medium based on voiceprint recognition

Publications (1)

Publication Number Publication Date
CN114387974A true CN114387974A (en) 2022-04-22

Family

ID=81195388

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111516627.9A Pending CN114387974A (en) 2021-12-13 2021-12-13 Case-crossing method, system, device and storage medium based on voiceprint recognition

Country Status (1)

Country Link
CN (1) CN114387974A (en)

Similar Documents

Publication Publication Date Title
US11294624B2 (en) System and method for clustering data
Shi et al. An end-to-end trainable neural network for image-based sequence recognition and its application to scene text recognition
KR101312770B1 (en) Information classification paradigm
AU2016273851B2 (en) Accurate tag relevance prediction for image search
Bensefia et al. Information Retrieval Based Writer Identification.
WO2021047341A1 (en) Text classification method, electronic device and computer-readable storage medium
US20080313140A1 (en) Method and Apparatus for Multi-Dimensional Content Search and Video Identification
CN103403704B (en) For the method and apparatus searching arest neighbors
CN112035599B (en) Query method and device based on vertical search, computer equipment and storage medium
CN107844533A (en) A kind of intelligent Answer System and analysis method
US8612371B1 (en) Computing device and method using associative pattern memory using recognition codes for input patterns
CN110134777B (en) Question duplication eliminating method and device, electronic equipment and computer readable storage medium
CN106528527A (en) Identification method and identification system for out of vocabularies
CN115115825B (en) Method, device, computer equipment and storage medium for detecting object in image
Le et al. Document retrieval based on logo spotting using key-point matching
JP4802176B2 (en) Pattern recognition apparatus, pattern recognition program, and pattern recognition method
CN111177388B (en) Processing method and computer equipment
KR100842216B1 (en) Automatic document classification method and apparatus for multiple category documents with plural associative classification rules extracted using association rule mining technique
Bart Parsing tables by probabilistic modeling of perceptual cues
CN114387974A (en) Case-crossing method, system, device and storage medium based on voiceprint recognition
Kotsifakos et al. Query-sensitive distance measure selection for time series nearest neighbor classification
CN113724779B (en) SNAREs protein identification method, system, storage medium and equipment based on machine learning technology
Srihari et al. Language independent word spotting in scanned documents
CN114638277A (en) Industrial component classification method based on decision tree
CN113220843A (en) Method, device, storage medium and equipment for determining information association relation

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination