Audio frequency separation method based on NMF non-negative matrix factorization

A technology of non-negative matrix decomposition and separation method, applied in speech analysis, speech recognition, instruments, etc., can solve the problem of no non-negative matrix decomposition, etc., to achieve the effect of efficient separation work

Inactive Publication Date: 2014-06-18
SHANGHAI 8D WORLD NETWORK SCI & TECH
View PDF0 Cites 19 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Therefore, there is currently no non-negative matrix factorization method for audio separation

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
  • Audio frequency separation method based on NMF non-negative matrix factorization
  • Audio frequency separation method based on NMF non-negative matrix factorization
  • Audio frequency separation method based on NMF non-negative matrix factorization

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035] Such as figure 1 , figure 2 and image 3 As shown, a kind of audio separation method based on NMF non-negative matrix decomposition of the present invention, comprises auxiliary music speech distinguishing module and NMF non-negative matrix decomposition module, auxiliary music speech distinguishing module mainly utilizes the method of machine learning, by extracting a large amount of The relevant audio features of speech music audio are used as training samples, and the recognition model is obtained by training through the SVM algorithm;

[0036] The NMF non-negative matrix decomposition module uses the NMF non-negative matrix decomposition method to iteratively decompose the original audio signal matrix until the decomposition results meet the requirements of the cost function and the auxiliary distinction module.

[0037] The described auxiliary music voice distinguishing module mainly has the following aspects:

[0038] a. It is necessary to collect a large numb...

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 belongs to the field of voice signal factorization, and specifically relates to an audio frequency separation method based on NMF non-negative matrix factorization. The audio frequency separation method comprises an auxiliary music voice distinguish module and an NMF non-negative matrix factorization module. The method can relatively well separate voice audio frequency and music audio frequency mixed in the audio frequency by introducing the NMF novel mathematic research result and combining the audio frequency characteristic difference of the voice audio frequency and the music audio frequency, thereby obtaining relatively clear music audio frequency and voice audio frequency. By combining the NMF method and machine learning algorithm, the separation work of the audio frequency can be completed efficiently.

Description

technical field [0001] The invention relates to the field of speech signal decomposition, in particular to an audio separation method based on NMF non-negative matrix decomposition. Background technique [0002] When a piece of audio is mixed with background music audio and speech audio, in order to obtain a single speech audio or music audio for corresponding encoding, information retrieval, etc., it is necessary to use the corresponding algorithm to separate the mixed audio into simple audio and music audio. Such an audio processing method is generally solved by using related technologies of signal separation. With the development of technology in the field of audio processing, the need for such technology has also spawned corresponding research results. [0003] When faced with the problem of separating such voice and music aliasing signals, it is mainly based on fundamental frequency analysis and spectrum decomposition algorithms. For example, Klapuri et al. proposed ...

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): G10L21/028G10L19/032G10L15/06
Inventor 王雷
Owner SHANGHAI 8D WORLD NETWORK SCI & TECH
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