Unlock instant, AI-driven research and patent intelligence for your innovation.

A method and system capable of training a recognition model according to the extraction frequency of the model

A technology for identifying models and frequencies, applied in speech analysis, instruments, etc., can solve the problem that the training corpus should not be too long

Active Publication Date: 2020-10-23
YUTOU TECH HANGZHOU
View PDF12 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in practical applications, such as voiceprint recognition applied to some smart devices for voice operations, it is required to have a high recognition accuracy, and it is also required that the training corpus should not be too long, so as to ensure better practicability. The technical solution of voiceprint recognition model establishment in the prior art is difficult to achieve the above purpose

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
  • A method and system capable of training a recognition model according to the extraction frequency of the model
  • A method and system capable of training a recognition model according to the extraction frequency of the model
  • A method and system capable of training a recognition model according to the extraction frequency of the model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0076] figure 1 It shows the implementation process of the method for training the recognition model according to the extraction frequency of the model provided by the first embodiment of the present invention. A plurality of clients and a server are provided, and the server is remotely connected to the plurality of clients respectively. The details are as follows:

[0077] In step S1, the client acquires an initial voice signal stream of a speaker.

[0078] In this embodiment, the method of training the recognition model according to the extraction frequency of the model may be used for an intelligent terminal in a private space such as an intelligent robot, so the initial voice signal flow can be the user performing voice chat or making a voice through the intelligent terminal The voice signal stream generated by instructions and the like may also be a voice signal stream obtained by means of recording or the like. Specifically, the above-mentioned method of training the re...

Embodiment 2

[0101] figure 2 The implementation process of the method for training the recognition model according to the extraction frequency of the model provided by the second embodiment of the present invention is shown, and the details are as follows:

[0102] In step S21, the client establishes a plurality of initial recognition models according to a plurality of preset sentence training samples.

[0103] Wherein, the initial recognition model is a recognition model established by calling the voiceprint registration algorithm interface according to the sentence training samples of the preset voice signal stream, and the initial recognition model is formed after the voiceprint registration process for a certain person or multiple people The registration process does not require the length of the training corpus or the flow sentence training samples of the speech signal. And because the method provided by the embodiment of the present invention can realize operations such as continuo...

Embodiment 3

[0125] image 3 It shows the structure of the system that can train the recognition model according to the extraction frequency of the model provided by the third embodiment of the present invention. The terminal provided by the third embodiment of the present invention can be used to implement the methods realized by the first to second embodiments of the present invention. For the convenience of description , only shows the parts related to the embodiment of the present invention, and the specific technical details are not disclosed, please refer to Embodiment 1 and Embodiment 2 of the present invention.

[0126] The system that can train the recognition model according to the extraction frequency of the model can be an intelligent terminal that is applied in a private space or a semi-open space and supports voice operations, such as an intelligent robot. In this embodiment, the recognition model can be trained according to the extraction frequency of the model. The system o...

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 discloses a method and system capable of training recognition models according to extraction frequencies of the models, and belongs to the technical field of voice recognition. The method capable of training the recognition models according to the extraction frequencies of the models adopts a mode of remote connection of a server and a client to perform data communication, can delete rarely-used initial recognition models in the client through comparison of the extraction frequencies of the initial recognition models, adopts sentence training samples in the server to update the rarely-used initial recognition models, thereby reducing the operation load of the client, improving work efficiency at the same time, and both relatively good practicability required by forming of recognition models when the method is applied to an ordinary intelligent terminal and the degree of accuracy required by voiceprint recognition can be considered.

Description

technical field [0001] The invention relates to the technical field of speech recognition, in particular to a method and system for training a recognition model according to the extraction frequency of the model. Background technique [0002] Voiceprint recognition is a recognition technology that utilizes human voice. Since there are certain differences in the vocal organs used by people when speaking, the voiceprint maps of any two voices are different, so voiceprints can be used to represent individual differences. Therefore, it is possible to characterize different individuals by establishing a recognition model, and then use the recognition model to identify different individuals. At present, there is a dilemma in the application of the recognition model, which is mainly reflected in the selection of the length of the training corpus. Generally speaking, the longer the voiceprint training corpus, the more accurate the feature model established, and the higher the recog...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G10L17/04G10L17/02
CPCG10L17/02G10L17/04
Inventor 祝铭明
Owner YUTOU TECH HANGZHOU