Robust Voice Activity Detection Feature Design Based on Spectral Kurtosis

In traditional VAD algorithms, High Order Statistics (HOS) is usually used in time domain and limited to white noise case. In this paper, a spectral domain HOS feature called spectral kurtosis is introduced, on the bases of which an essential exploring to the different characters between speech and noise in spectral domain is carried out. By the introducing of ldquotime delayrdquo and double thresholds method, an effective VAD algorithm based on spectral kurtosis is proposed. Experiment results show that the proposed feature and VAD algorithm based on it has better performance than other ones such as short-term energy, entropy, cepstral distance, etc, especially when the SNR and types of noise are time varying, so it has more applications in speech signal processing.