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

Voice dereverberation method and device

A technology of de-reverberation and speech, applied in speech analysis, instruments, etc., can solve the problems of inability to improve and improve the performance of speech communication system and ASR, large amount of calculation, and easy to be affected by noise.

Active Publication Date: 2019-07-05
CHINA ACAD OF TELECOMM TECH
View PDF11 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Embodiments of the present invention provide a method and device for voice reverberation to solve the problem of inability to improve and improve voice communication due to the inability to process voice signals in real time, the large amount of calculation, and the vulnerability to noise when offline WPE is used for voice processing. System and ASR performance

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
  • Voice dereverberation method and device
  • Voice dereverberation method and device
  • Voice dereverberation method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0279] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0280] At present, speech reverberation processing technology can be divided into three categories according to the principles followed by its algorithm processing, namely: sound source model-based reverberation technology, homomorphic transformation-based reverberation technology, and acoustic channel-based channel) inverse filtering and equalization de-reverberation technology.

[0281] Source model-based dereverberation techniques compute an estimate of a clean speech signal by exploiting a priori information about its structure and how it is distorted in the presence of room reverberation. Typical algorithms in this category are linear prediction (LP) residual enhancement, harmonic filtering, and dereverberation based on speech probability model...

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 provides a voice dereverberation method and a voice dereverberation device. The voice dereverberation method comprises the following steps: acquiring the frequency spectrum vector of each frame of voice signals captured by a microphone array; according to the frequency spectrum vector of each frame of voice signals, acquiring first frequency points with a detection result being a signal frame of voice and second frequency points with a detection result being a signal frame of non-voice; acquiring a first voice signal frequency spectrum vector of frequency spectrum at each frequency point in the first frequency points after dereverberation, and updating a predictive coefficient matrix of a weighing forecast error of each frequency point according to the first voice signal frequency spectrum vector; acquiring non-voice signal frequency spectrum vector of the frequency spectrum of each frequency point in the second frequency points after attenuation; and according to the first voice signal frequency spectrum vector and the non-voice signal frequency spectrum vector, acquiring voice signals after dereverberation. With the technical scheme, the real-time processing for thevoice signals is realized, the calculating amount is reduced, the influences of noise on the voice signals are reduced, and the performances of a voice communication system and an ARS are improved.

Description

technical field [0001] The invention relates to the technical field of speech signal processing, in particular to a method and device for reverberating speech. Background technique [0002] The reverberation effect will occur in the speech signal indoors, and the reverberated speech usually seriously reduces the speech intelligibility in the communication system and the correct speech recognition rate in the automatic speech recognition system (ASR). [0003] The existing de-reverberation technology has its own technical defects, especially when using offline WPE for voice processing, it cannot process voice signals in real time, has a large amount of calculation, is easily affected by noise, and cannot improve and enhance the voice communication system. and ASR performance. Contents of the invention [0004] Embodiments of the present invention provide a method and device for voice reverberation to solve the problem of inability to improve and improve voice communication...

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
IPC IPC(8): G10L21/0208G10L25/18
CPCG10L21/0208G10L25/18G10L2021/02082
Inventor 李佳芮梁民
Owner CHINA ACAD OF TELECOMM TECH
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More