复杂的噪声环境是语音识别系统在实际应用中性能下降的原因之一,识别预处理中的带噪端点检测作为关键技术,其性能的优劣某种程度上决定了识别率的高低。笔者提出了基于LPC美尔倒谱特征的带噪端点检测方法,对语音信号分高低频段分别提取LPC美尔倒谱特征分析,根据Mel倒谱距离判决,采用自适应噪声估计,实验结果表明,该方法计算效率较高,低信噪比下有较好的检测性能。 Complicated noise environment is one of the reasons why the performance of speech recognition system is degraded in practical application. As the key technology, the detection of noisy endpoint in recognition preprocessing determines the recognition rate to a certain extent. The author proposes a noise-free endpoint detection method based on the LPC cepstrum feature, and separately extracts the LPC cepstrum features of the speech signal in high and low frequency bands. According to Mel cepstrum distance decision, adaptive noise estimation is used. The experimental results show that, The method has high computational efficiency and good detection performance under low SNR.