Title:
VOICE ENHANCEMENT METHOD BASED ON VOICE PROCESSING AND RELATED DEVICE
Document Type and Number:
WIPO Patent Application WO/2020/224226
Kind Code:
A1
Abstract:
Provided are a voice enhancement method based on voice processing and a related device, wherein, the voice enhancement method based on voice processing comprises: acquiring original voice signals, sequentially inputting the original voice signals to a filter and a microphone array to obtain a fixed-beam microphone array of the original voice signals (S1); dividing the original voice signals into N sub-space voice signals according to the fixed-beam microphone array (S2); extracting a plurality of awakening words from each sub-space voice signal, and conducting framing and windowing processing (S3); calculating the confidence level of the corresponding sub-space voice signal according to the framed and windowed awakening words corresponding to each sub-space voice signal (S4); selecting the sub-space voice signal with the highest confidence level as the final output voice signal according to the calculation result of the confidence level (S5). The method disperses the directionality of noise, reduces the direction number of noise, extracts the sub-space voice signals with the high effective voice and weak noise, and thus achieves the voice enhancement effect.
Inventors:
ZHAO JIANPING (CN)
MA JUN (CN)
WANG SHAOJUN (CN)
MA JUN (CN)
WANG SHAOJUN (CN)
Application Number:
PCT/CN2019/118677
Publication Date:
November 12, 2020
Filing Date:
November 15, 2019
Export Citation:
Assignee:
PING AN TECH SHENZHEN CO LTD (CN)
International Classes:
G10L21/0216; G10L25/27
Foreign References:
CN110223708A | 2019-09-10 | |||
CN107785029A | 2018-03-09 | |||
CN106782504A | 2017-05-31 | |||
CN107424625A | 2017-12-01 | |||
US20090150146A1 | 2009-06-11 |
Attorney, Agent or Firm:
BEIJING JINGDA LAW FIRM (CN)
Download PDF: