Abstract: In this article, we utilizing twice sampling stochastic resonance, combined with twice sampling, realized the speech-denoising of weak speech signals. The traditional speech-denoising methods treat the noise as harmful interference signal, and eliminate the noise from the noise speech signal. But in low signal-to-noise(SNR) ratio environment, it will lost very much of the speech information, or make the waveform distortion. Stochastic resonance can transfer the noise energy to the weak signal, but classic SR is only realized under small parameters in terms of the theory of adiabatic elimination, which can’t using in speech-denoising. So our group using the technique of twice sampling realized the speech-denoising using stochastic resonance. The experiment result prove that this method can enhance the weak speech signals and improve initial SNR from -9 to -2.37, markedly enhance the SNR of output speech signals.
Keywords: stochastic resonance, twice sampling, speech denoising, signal-to-noise ratio.
Title: The Weak Speech-Denoising Base on Twice Sampling Stochastic Resonance
Author: Wenjian Cai
International Journal of Computer Science and Information Technology Research
ISSN 2348-1196 (print), ISSN 2348-120X (online)
Research Publish Journals
