论文部分内容阅读
为改善多通道助听器滤波器分解导致语音质量下降的问题,提出了一种基于听力图的非均匀余弦调制滤波器组的有效设计方法.首先根据线性迭代算法设计低通原型滤波器,然后根据原理公式设计得到均匀余弦调制滤波器组,最后根据听损患者听力图将变化平缓的频率范围对应的均匀滤波器组若干通道合并为一个通道,得到对应的非均匀余弦调制滤波器组.波形实验结果显示信号经非均匀滤波器组后的时域波形与语谱图无明显失真,差异较小.语音质量实验结果显示所提出的非均匀余弦调制滤波器组的主观语音质量评估值PESQ比传统滤波器组高出约35%,语音质量指数HASQI提高约40%.
To improve the speech quality degradation caused by multi-channel hearing aid filter decomposition, an effective design method of non-uniform cosine-modulated filter bank based on audiogram is proposed.First, the low-pass prototype filter is designed according to the linear iterative algorithm, and then according to the principle Formula is designed uniform cosine modulation filter bank, and finally according to hearing loss patients audiogram will change the frequency range of the smooth uniform frequency filter bank corresponding to a number of channels into a channel to obtain the corresponding non-uniform cosine modulation filter bank waveform results The time-domain waveform and the spectrogram of the display signal after the non-uniform filter group have no obvious distortion, and the difference is small.The experimental results of voice quality show that the subjective voice quality evaluation value PESQ of the proposed non-uniform cosine modulated filter bank is lower than the traditional filter About 35% higher than the device group, voice quality index HASQI increased by about 40%.