Unlock instant, AI-driven research and patent intelligence for your innovation.

Speech dereverberation method for joint beamforming and deep complex u-net networks

A de-reverberation and beam technology, which is applied in speech analysis, instruments, etc., can solve the problems of slow calculation speed, reduced de-reverberation effect, and unsatisfactory real-time application, so as to improve the signal-to-noise ratio and improve performance.

Active Publication Date: 2022-05-03
ZHEJIANG UNIV
View PDF16 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The multi-channel linear prediction method can achieve effective speech reverberation when the acoustic impulse response is unknown, but the disadvantage is that the calculation speed is slow and does not meet the needs of real-time applications
The common disadvantage of the beamforming method and the channel linear prediction method is that the effect of reverberation will be greatly reduced under the condition of low signal-to-noise ratio

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
  • Speech dereverberation method for joint beamforming and deep complex u-net networks
  • Speech dereverberation method for joint beamforming and deep complex u-net networks
  • Speech dereverberation method for joint beamforming and deep complex u-net networks

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0049] The specific embodiments of the present invention will be further described below in conjunction with the accompanying drawings.

[0050] like figure 1 As shown, the embodiment of the present invention provides a joint beamforming and deep complex U-Net network speech reverberation method, the specific implementation is as follows:

[0051] (1) Use the MVDR beamformer to preprocess the multi-channel voice collected by the microphone array to obtain the beamforming output Y bf ; The specific implementation is as follows:

[0052] Remember the weight vector of the MVDR beamformer The formula is as follows:

[0053]

[0054] in Represents the covariance matrix of the signal received by the microphone, Denotes the room impulse response corresponding to microphone q, (·) H Represents the transpose operation, f represents the frequency point;

[0055] Obtain the output signal Y after beamforming bf , the formula is as follows:

[0056]

[0057] Where X(t,f) i...

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 voice de-reverberation method for combined beamforming and deep complex U-Net network. The method includes: preprocessing the reverberant speech with a minimum variance distortionless response (MVDR) beamformer to suppress interference from non-target speech directions and improve the signal-to-noise ratio; predicting the desired speech using a deep complex U‑Net network The amplitude and phase spectrum of speech; the desired speech signal in the time domain is recovered by inverse short-time Fourier transform. The invention can be used to solve the problem of reverberation of speech in common indoor environments such as conference rooms, classrooms, living rooms, etc., enhance the speech signal received by intelligent interactive devices, and improve the accuracy of speech recognition and speech wake-up.

Description

technical field [0001] The present invention relates to a voice de-reverberation method, in particular to a voice de-reverberation method of combined beamforming and deep complex U-Net network. Background technique [0002] Speech is one of the most important and commonly used forms of exchanging information for humans. In recent years, with the development of computer science and pattern recognition technology, speech has become an important means of human-computer interaction. Due to reflections from room walls and other objects, the signal received by the microphone in an enclosed environment is a superposition of the direct wave and the reverberation. Reverberation destroys structures such as the envelope and harmonics of speech, resulting in reduced speech quality and clarity. In the presence of reverberation, the performance of automatic speech recognition systems can be significantly degraded. Therefore, it is more urgent to extract a relatively pure target speaker...

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): G10L21/0208G10L21/0216G10L25/30
CPCG10L21/0208G10L21/0216G10L25/30G10L2021/02082G10L2021/02166
Inventor 潘翔朱训谕
Owner ZHEJIANG UNIV