A new algorithm for voice activity detection based on wavelet transform
Xang Shaojun, Haitao Guo, Yin Fu-liang · 2005
In this paper, a new algorithm based on the wavelet transform (WT) is proposed for voice activity detection (VAD). It utilizes the difference of the spectral distribution between voice and noise. First, the algorithm performs the wavelet transform to the signal and decomposes it into subbands using the bandpass filtering feature of the WT; then it detects the voice in the signal by comparing the subband energy of the detail components between voice and noise. Computer simulation results are given to illustrate the effectiveness of the new VAD algorithm.