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

Multi-channel non-negative matrix factorization method and system based on frequency domain convolution transfer function

A non-negative matrix decomposition and transfer function technology, which is applied in speech analysis, complex mathematical operations, instruments, etc., can solve the problems of Fast-MNMF performance degradation and the non-existence of accurate joint diagonalization, so as to achieve good separation performance and improve separation performance, effect of increasing stats

Active Publication Date: 2022-03-22
INST OF ACOUSTICS CHINESE ACAD OF SCI
View PDF17 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the exact joint diagonalization of more than two non-negative definite Hermitian matrices does not exist, so the performance of Fast-MNMF degrades when separating more than two sound sources

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
  • Multi-channel non-negative matrix factorization method and system based on frequency domain convolution transfer function
  • Multi-channel non-negative matrix factorization method and system based on frequency domain convolution transfer function
  • Multi-channel non-negative matrix factorization method and system based on frequency domain convolution transfer function

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0025] The present invention will be further described in conjunction with accompanying drawing and example now.

[0026] Such as figure 1 As shown, the mixing process of the microphone array capturing the sound source and the unmixing process of the blind source separation algorithm to decompose the mixed signal are shown. The unmixing process can be regarded as the inverse process of the mixing process. During the mixing process, the direct sound waves of the sound source and the sound waves reflected by the walls of the room are recorded by the microphone at the same time. The reflected signal emitted by the wall of the room or other objects in the room and reaching the microphone is what we usually call the reverberation signal, and the reverberation signal has a great influence on the performance of the blind source separation algorithm. During the unmixing process, the blind source separation algorithm uses only the signal recorded by the microphone array to recover th...

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 technical field of blind source separation, and particularly relates to a multi-channel non-negative matrix factorization method and system based on a frequency domain convolution transfer function, and the method comprises the steps: carrying out the framing of a time domain observation signal collected by each channel of a microphone array, and carrying out the short-time Fourier transform, and obtaining a time-frequency domain observation signal; estimating the power spectrum density of each sound source based on a non-negative matrix factorization sound source model; estimating each order of demixing filter of each sound source based on the frequency domain convolution transfer function space model; constructing a Wiener filter under a mean square error criterion by using the obtained power spectral density and the demixing matrix, and filtering the observation signal of the time-frequency domain to obtain a separation signal of the time-frequency domain; and performing short-time inverse Fourier transform on the time-frequency domain separation signal, and synthesizing to obtain a time-domain separation signal.

Description

technical field [0001] The invention belongs to the technical field of blind source separation (BSS), and in particular relates to a multi-channel non-negative matrix decomposition method and system based on frequency-domain convolution transfer function. Background technique [0002] Blind source separation is a method of estimating each sound source signal using only the received signal of the microphone without prior information such as the sound source and the transfer function between the sound source and the microphone. Audio blind source separation has important applications in areas such as automatic speech recognition, automatic music transcription, and target speaker extraction in noisy environments. [0003] In fields such as biomedical signal or image processing, the observed signal is an instantaneous mixed model in the time domain. However, in the application field of audio signal processing, due to the existence of early reflections and late reverberation in ...

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): G10L21/0272G10L19/26G10L19/02G10L25/18G10L25/24G06F17/14G06F17/15
CPCG10L21/0272G10L19/26G10L19/02G10L25/18G10L25/24G06F17/14G06F17/15
Inventor 王泰辉
Owner INST OF ACOUSTICS CHINESE ACAD OF SCI
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

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com