Training method and apparatus of neural network model, and speech denoising method and device

A neural network model and training device technology, applied in the field of data processing, can solve the problems of poor versatility of denoising methods, and achieve the effect of improving versatility

Pending Publication Date: 2019-02-22
GREE ELECTRIC APPLIANCES INC
View PDF12 Cites 27 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0009] The embodiment of the present invention provides a neural network model training method, a voice denoising method and a device to solve the problem in the prior art that when different data to be denoised are input, the existing denoising method needs to be based on the specific conditions of the input data The technical problem of poor versatility of the denoising method due to manual adjustment

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 and apparatus of neural network model, and speech denoising method and device
  • Training method and apparatus of neural network model, and speech denoising method and device
  • Training method and apparatus of neural network model, and speech denoising method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0057] In order to solve the above technical problems, the general idea of ​​the technical solution in the embodiment of the present invention is as follows:

[0058] Provided are a neural network model training method, a voice denoising method and a device.

[0059] Specifically, for the training method and device of the neural network model:

[0060] Acquiring a voice sample set, wherein the voice sample set includes several voice samples;

[0061] Obtaining a scene noise sample set, wherein the scene noise sample set includes several scene noise samples;

[0062] Combining any speech sample in the speech sample set with any scene noise sample in the scene noise sample set until all speech samples in the speech sample set and all scene noise samples in the scene noise sample set are completed Combining, thereby obtaining several mixed sound samples and a mixed sound sample set comprising the several mixed sound samples;

[0063] extracting time-frequency features of each ...

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 training method and apparatus of a neural network model, and a speech denoising method and device. The training apparatus trains a neural network by performing a neural network training method to obtain a noise-separated neural network model; the speech denoising device performs a speech denoising method to obtain scene noise data or speech data in a tested speech sampleby using the noise-separated neural network model; and then the scene noise data or speech data in the tested speech sample are removed. Therefore, a technical problem that the existing denoising method needs to be manually adjusted according to the specific condition of the input data when different to-be-denoised data are inputted and thus the universality of the denoising method is poor in theprior art is solved.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a training method of a neural network model, a voice denoising method and a device. Background technique [0002] The current mainstream denoising methods are as follows: [0003] (1) Wavelet denoising, this method divides the frequency band of the signal into multiple levels, and then adaptively selects the corresponding frequency band to match the signal spectrum, so it has a good effect in retaining the subtle information of the data. [0004] (2) Based on the regularization method, by adding appropriate constraints in the noise reduction process, the ill-conditioned process can be transformed into a well-conditioned process. The diversification of noise sources and the difference of noise distribution lead to different types of noise, so the noise model will become complicated, and the solution state of the noise reduction process will appear ill-conditioned. Regularization-b...

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 Applications(China)
IPC IPC(8): G10L21/0208G06N3/04G10L15/06G10L21/0272G10L25/30
CPCG10L15/063G10L21/0208G10L21/0272G10L25/30G06N3/045
Inventor 刘欢陈彦宇马雅奇谭泽汉闫昊
Owner GREE ELECTRIC APPLIANCES INC
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap