Robust Endpoint Detection Algorithm of Chinese

In order to solve the problem of endpoint detection in presence of multi noises, this paper presents a robust algorithm of Chinese. Without any priori information of noise statistics, this approach employs the autocorrelation of low frequency coefficients in wavelet transform to detect the endpoint of voiced signal, and combines with the power spectral density of noisy speech to determine the commence moment of unvoiced signal. Simulation results indicate that this method has the perfect accuracy under very complex background noise.