Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Voice recognition method and voice recognition terminal

A speech recognition and terminal technology, applied in speech recognition, speech analysis, instruments, etc., can solve the problems of high cost and large amount of data, and achieve the effect of reducing input cost, reducing data to be processed, and reducing the amount of dialect information

Inactive Publication Date: 2019-07-19
CHINA UNITED NETWORK COMM GRP CO LTD
View PDF10 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The present invention at least partly solves the problem of large amount of data and high cost that the central server needs to process during dialect recognition in the existing speech recognition method, and provides a speech recognition method that reduces the amount of data and cost processed by the central server

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Voice recognition method and voice recognition terminal
  • Voice recognition method and voice recognition terminal
  • Voice recognition method and voice recognition terminal

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0027] Such as figure 1 As shown, this embodiment provides a method for speech recognition, a terminal based on speech recognition, the method includes:

[0028] S11. Determine the current dialect.

[0029] Wherein, that is to say, determine the dialect type of the current location, for example, if the terminal is in Guangdong province, the dialect type is Cantonese, and if the terminal is in Zhejiang province, the dialect type is Zhejiang dialect, etc.

[0030] S12. Obtain the voice information of the user, and under the preset dialect deep learning framework, according to the voice information and the preset general recognition model, train the dialect information model corresponding to the current dialect, and the preset general recognition model is used to recognize the voice information .

[0031] Among them, that is to say, the preset general recognition model can recognize voice information (such as Mandarin, etc.), but the preset general recognition model is not very...

Embodiment 2

[0038] Such as figure 2 As shown, this embodiment provides a method for speech recognition, a terminal based on speech recognition, the method includes:

[0039] S21. Receive a preset universal recognition model and a preset dialect deep learning framework from the central server.

[0040] For example, the terminal can download related software from the central server, and the software includes a preset general recognition model and a preset dialect deep learning framework. Specifically, the preset general recognition model can recognize speech information (such as Mandarin, etc.), but the preset general recognition model is not very accurate for the recognition of specific dialect speech information

[0041] S22. Determine the current dialect.

[0042] Wherein, that is to say, determine the dialect type of the current location, for example, if the terminal is in Guangdong province, the dialect type is Cantonese, and if the terminal is in Zhejiang province, the dialect type...

Embodiment 3

[0066] Such as image 3 As shown, this embodiment provides a speech recognition terminal, including:

[0067] The first acquisition module is used to determine the current dialect;

[0068] The model building module is used to obtain the voice information of the user. Under the preset dialect deep learning framework, according to the voice information and the preset general recognition model, the dialect information model corresponding to the current dialect is obtained through training. The preset general recognition model is used to for recognizing voice information.

[0069] The sending module is used to send the dialect information model to the central server, so that the central server can train to obtain a dialect recognition model corresponding to the current dialect, and the dialect recognition model is used to recognize the current dialect.

[0070] Preferably, the model building module includes:

[0071] The receiving sub-module is used to receive the user's voice...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention provides a voice recognition method and a voice recognition terminal and belongs to the technical field of voice recognition with an aim to at least partially solve problems of large amount of data and high cost that a center server needs to deal with when an existing voice recognition method is used for dialect recognition. The voice recognition method based on the voice recognitionterminal includes: determining current dialect; acquiring voice information of a user, acquiring a dialect information model of the corresponding current dialect by training according to the voice information and a preset general recognition model under the preset dialect depth learning framework, wherein the preset general recognition model is used for recognizing the voice information; sendingthe dialect information model to the center server for the training of the central server to acquire the dialect recognition model corresponding to the current dialect, wherein the dialect recognitionmodel is used for recognizing the current dialect.

Description

technical field [0001] The invention belongs to the technical field of voice recognition, and in particular relates to a voice recognition method and a terminal. Background technique [0002] With the continuous growth of user demands, the application of dialect speech recognition in electronic equipment is becoming more and more important. The existing dialect speech recognition method mainly sends the user's dialect speech information and geographical location to the central server multiple times through the user terminal, and then the central server continuously trains and analyzes according to the dialect speech information and geographical location of multiple users. Finally, a dialect speech recognition module is formed. [0003] However, there are many kinds of dialects in our country. If the dialect voice information in each region is sent to the central server for training and analysis many times, the data of the dialect voice information will be too concentrated i...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G10L15/22G10L17/04G10L15/00G10L15/06G10L17/18
CPCG10L15/005G10L15/063G10L15/22G10L17/04G10L17/18
Inventor 龙岳
Owner CHINA UNITED NETWORK COMM GRP CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products