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

Endpoint detection method at low SNR environment

An endpoint detection, low signal-to-noise ratio technology, applied in voice analysis, instruments, etc., can solve problems such as endpoint detection performance degradation

Inactive Publication Date: 2019-10-18
GUILIN UNIV OF ELECTRONIC TECH
View PDF2 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Aiming at the deficiencies in the prior art, the technical problem solved by the present invention is the problem that the performance of endpoint detection drops sharply in the environment 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
  • Endpoint detection method at low SNR environment
  • Endpoint detection method at low SNR environment
  • Endpoint detection method at low SNR environment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0076] Embodiment: The experiment uses M-Audio multi-channel audio equipment to collect voice data in a relatively quiet and open place. The speech content is three command words: "Xiaobai Xiaobai", "Turn on the speaker", "Xiaobai Xiaobai". The sampling frequency is 16kHz and the precision is 16bit. In the algorithm, the voice signal is added with a Hamming window for frame processing.

[0077] figure 1 Shows an endpoint detection method in a low SNR environment, by performing modulation domain spectral subtraction on noisy speech to obtain a balance between noise reduction and speech distortion, thereby improving speech quality, combined with power normalized cepstrum The distance between numbers is used for endpoint detection, including the following main steps:

[0078] (1) Carry out modulation domain spectrum subtraction to the noise-containing speech and compensate the speech after the phase is enhanced, the specific sub-steps are as follows:

[0079] (1) Assuming that...

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 endpoint detection method at a low SNR environment. By carrying out modulation domain spectral subtraction on voices with noises, balance between noise reduction and voice distortion can be obtained and voice quality is improved, and then endpoint detection can be carried out by combining with distance between power normalized cepstrum coefficients. The endpoint detection method includes the following main steps: (1) carrying out the modulation domain spectral subtraction on the voices with the noises and compensating for phase positions to obtain enhanced voices; and (2) extracting PNCC characteristic parameters of voice signals receiving the modulation domain spectral subtraction, and calculating cepstrum distance between PNCC cepstrum coefficients and noise cepstrum coefficients of each frame of signals; and using the cepstrum distance as a detection parameter, determining a threshold value according to the PNCC cepstrum distance of NIS frames, and then using a double threshold decision method to carry out the endpoint detection. The endpoint detection method effectively solves the problem that performance of the endpoint detection decreases sharply atthe low SNR environment, thus achieving higher detection accuracy in the environment of strong noise interference.

Description

technical field [0001] The invention relates to the technical field of speech signal processing, in particular to an endpoint detection method in a low signal-to-noise ratio environment. Background technique [0002] Endpoint Detection (Endpoint Detection, ED) usually refers to distinguishing speech signals and non-speech signals from background noise environments, and determining the start and end points of speech, also known as Voice Activity Detection (VAD). With the popularization of smart homes, the development of voice technology has also attracted much attention. People hope to use voice to control smart devices in more complex environments. Therefore, it is of practical value to study efficient voice control technology in low signal-to-noise ratio environments. . [0003] Endpoint detection is a commonly used front-end processing technology for voice signal processing. Accurate positioning of voice endpoints helps to eliminate interference from noise segments, enhan...

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): G10L25/78G10L25/81G10L25/84
CPCG10L25/78G10L25/81G10L25/84
Inventor 曾庆宁卜玉婷郑展恒
Owner GUILIN UNIV OF ELECTRONIC 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