Voice Activity Detection Algorithm Based on Mel-TEO in Noisy Environment

In order to solve the problem of the poor anti-noise performance, this paper proposes a new voice detection algorithm in noisy environment. On the basis of short-time TEO energy voice detection, increasing the judging section Mel Cepstral distance, we use two-stage decision mechanism with accurate judgment after rough judgment. First, we judge the voice endpoints preliminarily using Mel Cepstral distance which has a good anti-noise performance. Then, we judge the endpoints accurately using short-time TEO energy which reflects time-domain and voice formant characteristics of the speech signal. Experiments show that this algorithm has better capability in low SNR, compared with the traditional double thresholds and short-time TEO energy, without increasing the computational complexity and improving the accuracy of the detection system.