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

Unsupervised noise estimation and speech enhancement method based on separable deep automatic encoding technology

A technology of automatic coding and speech enhancement, applied in speech analysis, instruments, etc., can solve problems such as influence effect and mismatch

Active Publication Date: 2015-11-04
PLA UNIV OF SCI & TECH
View PDF5 Cites 21 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in the case of unknown noise, or the characteristics of the unknown noise are very different from the known noise, there will be a mismatch problem, which will affect its effect

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
  • Unsupervised noise estimation and speech enhancement method based on separable deep automatic encoding technology
  • Unsupervised noise estimation and speech enhancement method based on separable deep automatic encoding technology
  • Unsupervised noise estimation and speech enhancement method based on separable deep automatic encoding technology

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0090] The application principle of the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0091] combine figure 1 , the implementation process of the unsupervised noise estimation and speech enhancement method based on the separable deep automatic coding technology of the present invention is as follows.

[0092] S101, randomly select 500 sentences of speech from different genders and different speakers from the English classic database TIMIT, sample them at 8kHz, and use a frame shift of 64ms with a window length of 8ms as the parameter frame, and then do a fast Fu of 512 points Liye transform, after taking the modulus, extract their amplitude spectrum S;

[0093] S102, then implement non-negative matrix decomposition to S, and train a non-negative speech dictionary D capable of representing speech signals, wherein the size of the dictionary is the number of basis functions selected as 2000;

[0094] ...

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 an unsupervised noise estimation and speech enhancement method based on a separable deep automatic encoding technology. The method comprises a step of prior processing and a step of speech enhancement on unknown noise pollution. The method can be flexibly applied to various speech processing scenes. The method is not constrained by a language of a speech content, changes of a speaker, and a kind of noise. Compared with classical stationarity assumption-based spectral estimation algorithm SS (Spectrum Subtraction) and MMSE (Minimum Mean Square Error), the method does not rely on stationarity assumption and can accurately estimate spectrums for stationary or abrupt noise. Compared with algorithm based on a hidden Markov model and a linear prediction coefficient, the method does not need to specify the type of the processed non-stationary noise. Compared with a noise estimation method based on a low rank structure, noise in the method does not need to have a low-rank repeated structure.

Description

technical field [0001] The invention belongs to the technical field of speech signal processing, in particular to an unsupervised noise estimation and speech enhancement method based on separable deep automatic coding technology. Background technique [0002] Speech enhancement is of great significance both for improving the auditory effect of the speech signal and as a front-end processing to improve the performance of the speech recognizer. The core issue of speech enhancement is the separation of speech noise. The ideal speech enhancement technology needs to be able to achieve good results under the premise of unknown noise. For this reason, a key problem that needs to be solved in speech enhancement is the problem of noise estimation. In order to estimate the noise spectrum, some classical algorithms have been proposed, such as Spectrum Subtraction (SS), Minimum Mean Square Error (MMSE), etc., and have been widely used in voice communication. However, these methods are...

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): G10L21/0208G10L19/008
Inventor 孙蒙李轶南张雄伟王艺敏邹霞贾冲李莉
Owner PLA UNIV OF SCI & TECH
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