听觉模型用于噪声环境下语音识别的研究

Auditory-model-based Speech Recognition under Noisy Environment

在吸收听觉生理学研究成果基础上，建立了一个模拟外围听觉系统和部分中枢听觉神经系统功能的听觉模型．以该模型作为前端处理来提取特征参数，后处理采用模式匹配的方法．进一步研究了听觉模型用于语音识别的抗噪声问题，根据听觉生理特点提出了几个抗噪声方案．实验结果表明，基于文中的听觉模型和所提抗噪声方案的语音识别系统，在噪声环境下具有很好的鲁棒性（Ｒｏｂｕｓｔｎｅｓ）．

In this paper,on the basis of absorption of the achievements of the resarch on auditory physiology,an auditory model simulating the peripheral auditory system and part of the central auditory system is set up.With the auditory model as the front-end to extract character parameters,and the module matching method as the back-end,the problem of noise robustness of the speech recognition system is also studied and several solutions are proposed and implemented.The result of experiments shows that the system based on the given model and noise-resisting methods is robust in noisy environment.