Abstract: An auditory model has been presented through the concept of phonetic scene analysis in this paper.The mixed speech signals in scene have been separated through the improved FastICA algorithm,and identification of speech signals has been accomplished based on the combination of optimized time-frequency distribution and distance measure.The detailed experiment results has　been presented to illustrate the effectiveness of the new auditory model.

摘要： 从语音场景分析的角度提出了一种听觉模型。文中采用独立分量分析中的FastICA算法对场景中的混叠语音进行分离,完成了听觉分辨功能。对于听觉功能中的说话人验证问题,结合优化时频分布和距离测度方法完成了验证过程。通过对实验结果的详细评估说明提出的模型具有较好的处理效果,很好地模拟了人类听觉功能,误判率仅为2.4%。