论文部分内容阅读
为了在低信噪比和复杂噪声环境下检测汉语浊语音,根据浊语音谐波结构特性,提出了一种鲁棒的浊语音检测方法。通过改进的谱跟踪算法,得到能表征浊语音谐波特性的一簇谱线;从谱线簇中提取谐波特征作为汉语浊语音检测的依据。在不同信噪比和不同噪声环境下的浊语音检测对比实验中全面优于传统方法,在0 dB信噪比时正识率高于传统方法约30%。实验结果表明,该方法在低信噪比和非平稳复杂噪声环境下都具有较好的浊语音检测效果。
In order to detect turbid speech in Chinese under low SNR and complex noise conditions, a robust turbid speech detection method is proposed based on the structural characteristics of voiced speech harmonics. By using the improved spectral tracking algorithm, a cluster of spectral lines can be obtained which can characterize the voicing characteristics of voiced speech. Harmonic features are extracted from the spectral lines as the basis for Chinese voiced speech detection. Compared with the traditional method, the correct detection rate of turbid speech in different SNR and different noisy environments is better than that of the traditional method at 0 dB SNR. Experimental results show that this method has a good speech detection performance under low SNR and non-stationary complex noise conditions.