基于听觉事件检测的汉语语音声韵切分

Segmentation of Chinese initials and finals based on auditory event detection

摘要: 提出了一种基于听觉事件检测的汉语声韵母切分方法。该方法首先使用耳蜗滤波器组对语音进行滤波,然后在每个频带上检测对应于能量突变的听觉事件,最后在不同频率范围对听觉事件进行融合以确定声韵母边界。实验结果表明,对8 kHz采样的干净语音切分准确率可达到88.9%;信噪比10 dB的语音切分准确率可达到82.9%以上。

Abstract: This paper presents a segmentation method of Chinese initials and finals based on the detection of auditory events.According to this method,the voice should first of all be filtered by using the cochlear filter bank,and then the auditory events corresponding to energy mutation in each band are detected.Finally,the auditory events are integrated in different frequency ranges respectively to determine the boundaries of Chinese initials and finals.The experimental results show that with 8 kHz sampling frequency,the accuracy is 88.9% for clean speech and above 82.9% for noisy speech with the SNR of 10 dB.