An SAD algorithm based on SGMM and phoneme combination

Speech activity detection (SAD) is the key preprocess of speech application. This paper proposed a subspace Gaussian mixture model (SGMM) and phoneme combination based SAD algorithm. This algorithm is efficient, small and can utilize speech recognition corpus directly. Results indicate that, compared with the baseline, our proposed method results in an absolute improvement of 4.9% frame error rate and 10% average hit rate, respectively. Our approach finally achieves a frame error rate of 5.1% and an average hit rate of 91.5%. The model size is just 809.5K.