一种应用于语音识别的端点检测改进方法

语音端点检测是语音识别过程中的重要的一个环节,为了提高在强背景噪声条件下语音端点检测的性能,提出了一种将维纳滤波和改进的多子带熵相结合的方法.不仅有效地减少了背景噪声,而且大大提高了语音端点检测的准确性和稳健性.仿真实验表明,该方法计算简单,可靠信高,在较低的信噪比下仍能比较准确的检测到语音信号的端点.

An Improved Speech Endpoint Detection Method Based on Spectral Entropy

Speech endpoint was an important segment in the speech preprocess.To improve the robustness of the vad,we proposed the logarithm that combines the weinner filter with the multi-band entropy.This algorithm not only reduces the background noise,but also improves the precision and robustness of the endpoint detection algorithm.The simulation experiment demonstrate that the algorithm is simply and reliable.It has a good performance even in low SNR environments.