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

Training method of voice wake-up model, detection method and device of voice false wake-up

A technology of voice wake-up and training method, applied in voice analysis, voice recognition, instruments, etc., can solve the problem of high cost

Active Publication Date: 2022-08-02
深圳市友杰智新科技有限公司
View PDF9 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The main purpose of this application is to provide a training method for a voice wake-up model, a detection method and equipment for voice false wake-up, aiming at solving the disadvantages of retraining the voice recognition model in order to solve the new similar sound false wake-up, and the high cost

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
  • Training method of voice wake-up model, detection method and device of voice false wake-up
  • Training method of voice wake-up model, detection method and device of voice false wake-up
  • Training method of voice wake-up model, detection method and device of voice false wake-up

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0034] In order to make the objectives, technical solutions and advantages of the present application more clear, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application.

[0035] refer to figure 1 , an embodiment of the present application provides a training method for a voice wake-up model, including:

[0036] S1: obtain training corpus, the training corpus includes positive sample corpus and negative sample corpus, and the negative sample corpus includes basic sample corpus and difficult sample corpus, and the difficult sample corpus is a similar phonetic vocabulary corpus of a preset wake-up word;

[0037] S2: Input the training corpus into a neural network for model training, wherein the neural network includes a basic network, a classificat...

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 present application provides a method for training a voice wake-up model, a method and device for detecting false wake-up of voice. In the model training stage, the basic network and the classification layer are the recognition network structures of the existing voice wake-up model, and the embedding layer is a new The plug-in is added to the original identification network structure, so there is no need to change the frame of the original identification network structure, and it is easy to expand. In the model application stage, the basic wake-up accuracy is guaranteed by the first recognition result output by the basic network and the classification layer; on this basis, the vector between the second recognition result output by the basic network and the embedding layer and the wake-up word feature vector is added. Distance judgment, so as to identify whether the similarity between the current recognition result of the voice wake-up model and the preset wake-up word reaches the preset condition, thereby improving the overall recognition accuracy of the voice wake-up model for similar sounds, and effectively reducing the false wake-up rate.

Description

technical field [0001] The present application relates to the technical field of speech recognition, and in particular, to a method for training a speech wake-up model, a method and device for detecting false wake-up by speech. Background technique [0002] When the wake word and command word recognition technology is applied, it can monitor the user's voice in real time, and give corresponding feedback when a specific word is detected (for example, wake up the device when a wake word is detected). In practical applications, similar sounds, similar sounds are often misidentified and awakened by mistake, that is, when the pronunciation of the user's voice is close to the wake-up word, there is a certain probability that it will be misrecognized and the device will be awakened by mistake. The wake-up situation will bring a very bad user experience to the user. The existing processing for the false awakening of similar sounds usually adds negative samples to the training data ...

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): G10L15/06G10L15/10G10L15/16G10L15/22
Inventor 李杰王广新杨汉丹
Owner 深圳市友杰智新科技有限公司
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