Speech enhancement and detection method suitable for actual communication condition

A technology for speech enhancement and communication conditions, applied in speech analysis, instruments, etc., can solve problems such as large noise redundancy, low readability, and poor speech signal quality

Pending Publication Date: 2022-04-05
ARMY ENG UNIV OF PLA
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0009] The invention provides a voice enhancement and detection method suitable for actual communication conditions, aiming at problems such as poor quality of some voice signals, low readability, and large noise redundancy under actual communication conditions, using voice enhancement and voice endpoint detection technology for processing

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 enhancement and detection method suitable for actual communication condition
  • Speech enhancement and detection method suitable for actual communication condition
  • Speech enhancement and detection method suitable for actual communication condition

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035] A speech enhancement and detection method suitable for use in actual communication conditions, including the following steps:

[0036] Step 1: Enhance Pre-pretreatment: All of the speech data is required before use, respectively, and the voice signal is subjected to the voice signal, fragmentation, and pre-weight. Among them, the resampling refers to all speech signals for all speech signals at 16kHz sampling rate. Voice signals are processed in terms of speech frames, whether all speech signals are used in units of speech frames, so that all speech signals are framed by 8192 sampling points per frame, and the frame movement of the training phase is 50%, and the test Stage frame transfer set to 100%. Since the power profile of the speech signal is reduced with the increase in the frequency, most energy is concentrated in the low frequency portion, while the energy is relatively small, in order to improve the resolution of the high frequency portion of the speech signal, add...

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 and detection method suitable for an actual communication condition comprises the following steps: preprocessing an actual communication signal, inputting the preprocessed actual communication signal into a trained SEDWGAN-div model for speech enhancement processing, and outputting an enhanced actual communication signal; carrying out secondary noise reduction on the enhanced actual communication signal by using a parameter-adaptive multi-window spectrum estimation spectral subtraction method, extracting an adaptive sub-band logarithmic energy entropy product characteristic parameter of the signal, carrying out voice detection by using a dynamic threshold double-threshold detection method by taking the voice characteristic parameter as a threshold value, and outputting voice endpoint information; and the enhanced actual communication signal is screened and segmented according to output voice endpoint information, a voice segment carrying voice information is reserved, a non-voice redundant segment in a channel idle state is removed, and finally an enhanced actual communication voice segment is output. In order to solve the problems of poor quality, low readability, large noise redundancy and the like of part of voice signals under actual communication conditions, voice enhancement and voice endpoint detection technologies are used for processing.

Description

Technical field [0001] The present invention relates to the field of speech enhancement and speech endpoint detection processing, and more particularly to a speech enhancement and detection method suitable for use in actual communication conditions. Background technique [0002] The history of voice enhancement originated in Bell Labs in the beginning of the last century. In order to improve the communication quality of the telephone, the researchers have studied a lot of research in the direction of signaling, and more decades, more researchers conduct voice enhancement technology A more in-depth study, divide the voice enhancement into two phases according to the different enhanced models: there is no supervision and enhancement phase and the supervision and enhancement phase. [0003] There is no supervision and enhancement phase, and it is often referred to as a traditional voice enhancement phase. The so-called unison is not required to use large data to prepare an offline t...

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): G10L21/0232G10L25/18G10L25/27G10L25/60G10L25/78
Inventor 张洪德韩鑫怡张晓克王月磊杨琬田树林陈嘉成
Owner ARMY ENG UNIV OF PLA
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