Endpiont Detection in Noisy Speech Signal Using Teager Energy Entropy

Abstract

In this paper, we propose a new method using teager energy operator and entropy to solve endpoint detection problem in noisy environment. With the teager energy operator, it is sensitive on AM and FM signal and noise suppression capability on noisy speech signal, calculate teager energy of noisy speech signal. According to the different teager energy probability distribution between noise and speech signal, teager energy entropy is different. Set two soft thresholds of four states to detect the endpoint of noisy speech signal. The simulation shows that the method has good effect of endpoint detection in low SNR conditions, the simulation results show that teager energy entropy of speech signal endpoint detection in noisy environments is feasible and effective, and improves the reliability of endpoint detection.