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

A speech enhancement method based on deep learning assisted rls filter processing

A filter processing and voice enhancement technology, which is applied in voice analysis, instruments, etc., can solve the problems of reducing microphone beam signal processing power and enhancing voice signals, so as to improve voice recognition rate and human-computer interaction experience, and enhance voice signals , the effect of reducing noise residue

Active Publication Date: 2021-12-17
成都启英泰伦科技有限公司
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the defects in the prior art, the present invention discloses a speech enhancement method based on deep learning assisted RLS filter processing, which effectively reduces the computing power of microphone beam signal processing, and can reduce the output without increasing distortion Noise residue in the signal, enhance the speech signal, thereby improving the speech recognition rate

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
  • A speech enhancement method based on deep learning assisted rls filter processing
  • A speech enhancement method based on deep learning assisted rls filter processing
  • A speech enhancement method based on deep learning assisted rls filter processing

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0021] Specific embodiments of the present invention will be further described in detail below.

[0022] The speech enhancement method based on deep learning assisted RLS filter processing of the present invention, comprises the following steps:

[0023] S1. The beamforming method of generalized sidelobe cancellation is used to process the microphone array voice signal y(l,k) to obtain the fixed beamforming output signal y s (l,k) and noise reference signal u(l,k); l, k denote time and frequency index respectively;

[0024] S2. Randomly extract the feature signal of any microphone signal in the microphone array and send it to the GRU-Mask network to calculate the masking value mask (l, k) of the original microphone signal;

[0025] S3. Compare the masking value mask(l,k) output by the GRU-Mask network with the noise threshold thred:

[0026] When mask(l,k)0 (l,k) for a fixed beamforming output signal y s (l,k) do filtering processing, the processed signal is used as the fin...

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

A speech enhancement method based on deep learning-assisted RLS filtering processing, comprising the following steps: S1. Use a beamforming method of generalized sidelobe cancellation to process speech signals to obtain a fixed beamforming output signal and a noise reference signal; S2. Randomize Extract the characteristic signal of any microphone signal in the microphone array and send it to the GRU-Mask network to calculate the masking value of the original microphone signal S3. Compare the masking value output by the network with the noise threshold, calculate the noise canceller, and use the noise canceller to perform noise eliminate. The present invention only uses the RLS algorithm to filter the signal dominated by the noise component, which effectively reduces the computing power of the microphone beam signal processing, and can reduce the noise residue in the output signal without increasing the distortion, so as to enhance the speech signal and improve the The purpose of speech recognition rate and human-computer interaction experience.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence and relates to speech recognition, in particular to a speech enhancement method based on deep learning-assisted RLS filter processing. Background technique [0002] With the wide application of voice interaction technology, the traditional single-microphone voice enhancement method can no longer meet the needs of voice quality in interactive technology. For example, in far-field environments or noisy environments, the information captured by single-microphone methods is limited and the noise reduction performance is limited. At this time, using the microphone array signal can effectively use the directional information of the voice signal to capture the voice signal in the beam, suppress the signals of other beams, and obtain a better noise reduction effect. [0003] As one of the classic beamforming algorithms, the General Sidelobe Canceller (GSC) method is widely used. However,...

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): G10L21/0208G10L21/0216
CPCG10L21/0208G10L21/0216G10L2021/02166
Inventor 万东琴胡岸刘文通曾帆
Owner 成都启英泰伦科技有限公司
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