Multichannel speech enhancement method based on time-frequency domain binary mask

A binary mask, time-frequency domain technology, applied in speech analysis, instrumentation, etc., can solve problems such as estimation distortion and speech enhancement effect decline

Active Publication Date: 2020-10-23
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF6 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0011] In the above scheme, when the SNR of the received signal continues to be high or low, the algorithm’s estimation of beamforming related parameters is severely distorted, resulting in a decline in the effect of speech enhancement

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
  • Multichannel speech enhancement method based on time-frequency domain binary mask
  • Multichannel speech enhancement method based on time-frequency domain binary mask
  • Multichannel speech enhancement method based on time-frequency domain binary mask

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0042] The basic idea of ​​the present invention is to create a new beamforming parameter estimation method by constructing a binary mask estimation based on the estimated value of the time-frequency domain speech existence probability, and classify the time-frequency components of the signal by using the binary mask estimation , to eliminate the influence of the noise part on the beamforming as much as possible.

[0043] Example steps such as figure 1 Shown:

[0044] Step 1. Generate convolutional neural network (CNN) input features according to the speech signal data, and estimate the probability of speech existence.

[0045] Suppose the received signal in the time-frequency domain is:

[0046] x i (f k ) = a i (f k , θ)·S i (f k )+N i (f k )

[0047] where S i (f k ) is the i-th frame frequency f k The sound source signal component of a i (f k , θ)∈C M×1 represents the array pair f k Steering vector of the frequency signal, N i (f k )∈C M×1 It is zero-m...

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 relates to a multichannel speech enhancement method based on a time-frequency domain binary mask for receiving speech signals by an array. A binary mask is calculated by utilizing a network model to output speech existence probability estimation, and classification of a signal time-frequency domain and corresponding beamforming parameter estimation are realized through the binary mask, so that a better speech enhancement effect is obtained. The method comprises the following steps: firstly, performing time-frequency domain speech existence probability estimation on an array receiving signal by utilizing a network model, and then calculating a threshold value by utilizing an estimation result and the receiving signal, thereby calculating binary mask estimation and beam formingrelated parameter estimation so as to realize multi-channel speech enhancement. Compared with an existing array received signal speech enhancement algorithm, the method has a higher output signal-to-noise ratio and a higher subjective speech quality assessment PESQ score.

Description

technical field [0001] The invention belongs to the beam forming technology, in particular to the multi-channel language enhancement technology of time-frequency domain binary mask estimation. [0002] technical background [0003] With the research and development of pattern recognition and machine learning, some methods have been used for reference in the field of speech enhancement, and a series of speech enhancement algorithms combining machine learning and multi-channel speech enhancement have emerged. Compared with the traditional multi-channel speech enhancement algorithm, these algorithms estimate the specific mask of the received signal through the machine learning model, and then estimate the parameters related to beamforming more accurately, which can avoid the spatial distribution of the microphone array and the direction of the target. a priori assumptions for better speech enhancement performance. However, there are many types of machine learning models, and th...

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): G10L21/0216G10L21/0224G10L21/0232
CPCG10L21/0216G10L21/0224G10L21/0232G10L2021/02166
Inventor 江家麒
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA
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