【摘要】 为解决低信噪比环境下端点检测性能急剧下降的问题,研究一种端点检测的方法。利用调制域谱减法对带噪信号降噪,提高信噪比后进行经验模式分解(empirical mode decomposition, EMD),选择合适的imf(intrinsic mode function)分量重构信号;进行EMD后计算每阶imf分量的Teager能量算子,结合对数能量和相关函数法对语音信号进行端点检测。实验结果表明,该方法在不同噪声的低信噪比环境下检测正确率较高,具有一定的稳健性。更多还原

【Abstract】 To solve the problem of the sharp decline of voice activity detection performance in the low signal-to-noise ratio(SNR) environment, an improved endpoint detection method was proposed. The noisy speech was enhanced by modulation domain spectral subtraction. Some IMFs components that could be used in endpoint detection were selected to reconstruct the signal by comparing means and variances, and the Teager energy operator of each imf was calculated after the second EMD. The logarithmic energy and correlation were combined to detect the endpoints of speech signals. Experimental results show that the accuracy of the proposed method is greatly improved and it has better robustness to different types of noise in the low SNR environment.更多还原