摘要
根据人耳能从噪声中提取有用信息的听觉特征，并结合语音信号的基本特征，提出并研究了一个适合于语音增强的听觉模型；实验结果表明，这个方法不仅在提高语音信噪比方面，而且在减小语音失真度方面均有较好的改善。
This paper studies and simulates the hearing evidence that human auditory system can select some useful information from noisy disturbance. Based on some hearing evidence, we find the relation between hearing response and hearing threshold. Human auditory system adjusts threshold value of hearing response according to the useful sound information that we want to obtain. It is assummed that the auditory system fetched and recognized speech from background noise or disturbace just because of the spectrum of the dynamic threshold value. Thus an auditory model of speech enhancement is presented. Base on this model, some experiments about niosy speech are presented. The results show that this method of speech enhancement yields lower distortion as well as higher signal to noise (S/N) ratio than conventional methods.
出处
《南京大学学报（自然科学版）》
CSCD
1996年第2期218-223,共6页
Journal of Nanjing University（Natural Science）
基金
国家自然科学基金
