【摘要】 基于与头相关传递函数HRTFs在不同频率区间上频谱特征重要性程度存在着差异的判断,结合空间听觉的研究成果,应用多种群并行进化、参数实数编码等改进措施的遗传算法GA进行了HRTFs的零极点模型逼近.实验结果表明,GA在适合人耳听觉感知特性的对数幅度谱误差准则下,逼近效果较传统的Prony和Yule-W alker方法分别获得了平均39%和46%的改善.更多还原

【Abstract】 Based on the judgement of HRTFs’(head-related transfer functions,HRTFs) different importance for sound source’s recognition in different spectrum spans,and associated with some perspectives of psychoacoustics in spatial hearing,multiple demes’ parallel and real-valued coding genetic algorithm(GA) is applied to the approximation of the HRTFs’ zero-pole model.Using the logarithmic magnitude’s error criterion for human auditory sense,the results show that,on an average,the performance of GA is 39% better than that of the traditional Prony method,and 46% than that of the Yule-Walker algorithm.更多还原