一种强噪声环境下的语音端点检测算法

Speech Endpoint Detection Algorithm Based on Strong Noise Environment

语音识别中语音信号端点检测的准确度起着至关重要的作用,传统的频带方差算法在强背景噪声环境下性能下降。提出一种改进算法,在保证算法复杂度较低的条件下,大大提高了检测的准确度。仿真证明,该算法在低信噪比条件下也具有较强的鲁棒性。

The accuracy of speech recognition in speech signal endpoint detection plays an important role. Traditional frequency band variance algorithm gives a bad performance in strong noise environment. In ensuring a lower computational complexity con- ditions, Put forward an improved algorithm, which can greatly improve the accuracy of detection. The simulation results show that the algorithm has strong robustness under low SNR condition.