In this article we present speech detection systems based on Daubechie, Coiflet and Symlet wavelet transforms respectively. For each a selection of the most eligible levels of signal decomposition for the corrupted speech detection problem was made. Using those levels the distinction between noise and corrupted signal can be amplified as far as 100 times. Tests were accomplished using a set of Slovak words artificially noised to several SNR by white WSS noise.
Speech detection in the noisy environment using wavelet transform
2003-01-01
301050 byte
Conference paper
Electronic Resource
English
Speech Detection in the Noisy Environment using Wavelet Transform
British Library Conference Proceedings | 2003
|Speech estimation using body-conducted speech microphone in noisy environment
Tema Archive | 2009
|Robust Speech Detection for Noisy Environments
Online Contents | 2011
|Robust speech detection for noisy environments
IEEE | 2011
|