Paper Info
Title
Harmonic-Based Robust Voice Activity Detection For Enhanced Low Snr Noisy Speech Recognition System
Abstract
This paper describes a novel harmonic-based robust voice activity detection (H-RVAD) method with harmonic spectral local peak (HSLP) feature. HSLP is extracted by spectral amplitude analysis between the adjacent formants, and such characteristic can be used to identify and verify audio stream containing meaningful human speech accurately in low SNR environment. And, an enhanced low SNR noisy speech recognition system framework with wakeup module, speech recognition module and confirmation module is proposed. Users can determine or reject the system feedback while a recognition result was given in the framework, to prevent any chance that the voiced noise misleads the recognition result. The H-RVAD method is evaluated by the AURORA2 corpus in eight types of noise and three SNR levels and increased overall average performance from 49% to 2090. In home noise, the performance of H-RVAD method can be performed from 49% to 1490 sentence recognition rate in average.
Year
DOI
Venue
2016
10.1587/transfun.E99.A.1928
IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES
Keywords
Field
DocType
robust voice activity detection, harmonic spectral local peak, low SNR noisy speech recognition
Pattern recognition,Voice activity detection,Harmonic,Speech recognition,Artificial intelligence,Mathematics
Journal
Volume
Issue
ISSN
E99A
11
0916-8508
Citations 
PageRank 
References 
0
0.34
18
Authors
3
Name
Order
Citations
PageRank
Po-Yi Shih1365.61
Po-chuan Lin2475.73
Jhing-fa Wang3982114.31