为了提高噪声环境下语音端点检测的鲁棒性,提出了一种结合多子带能量特征和最优化边缘检测判决准则的算法。该算法的突出优点在于:在不同信噪比情况下,其端点检测滤波器的输出基本不变,从而避免了门限调整所带来的困难。实验结果表明,这种算法在多种噪声环境下都能够达到较好的语音检出效果。这种算法克服了传统语音端点检测以短时能量、基频、过零率等作为检测特征时,需要动态调整门限且在低信噪比情况下鲁棒性较差的缺点。 In order to improve the robustness of speech endpoint detection in noisy environments, an algorithm combining multi-subband energy features and the optimal edge detection decision criterion is proposed. The outstanding advantage of this algorithm is that the output of its endpoint detection filter is basically unchanged under different signal-to-noise ratios, thus avoiding the difficulties caused by the threshold adjustment. Experimental results show that this algorithm can achieve better speech detection under a variety of noise conditions. This algorithm overcomes the shortcomings of traditional speech endpoint detection, such as short-time energy, fundamental frequency, zero-crossing rate, etc., which need to dynamically adjust the threshold and have poor robustness under low signal-to-noise ratio.