Based on the measurements from 52 Chinese subjects (26 males and 26 females), a high-spatial-resolution head-related transfer function (HRTF) database with corre- sponding anthropometric parameters is established. By ...Based on the measurements from 52 Chinese subjects (26 males and 26 females), a high-spatial-resolution head-related transfer function (HRTF) database with corre- sponding anthropometric parameters is established. By using the database, cues relating to sound source localization, including interaural time difference (ITD), interaural level difference (ILD), and spectral features introduced by pinna, are analyzed. Moreover, the statistical relationship between ITD and anthropometric parameters is estimated. It is proved that the mean values of maximum ITD for male and female are significantly different, so are those for Chinese and western sub- jects. The difference in ITD is due to the difference in individual anthropometric parameters. It is further proved that the spectral features introduced by pinna strongly depend on individual; while at high frequencies (f≥ 5.5 kHz), HRTFs are left-right asymmetric. This work is instructive and helpful for the research on bin- aural hearing and applications on virtual auditory in future.展开更多
个性化的头相关传输函数(head-related transfer function,HRTF)可以有效改善空间音频质量。针对个性化HRTF难以精确获得的问题,提出了一种基于层级集成的个性化空间音频生成方法。该方法通过三个模型逐层建立个性化HRTF中的定位信息。...个性化的头相关传输函数(head-related transfer function,HRTF)可以有效改善空间音频质量。针对个性化HRTF难以精确获得的问题,提出了一种基于层级集成的个性化空间音频生成方法。该方法通过三个模型逐层建立个性化HRTF中的定位信息。首先,采用高斯混合模型建立用户无关的共用模型。然后,采用自编码器获得与用户有关的HRTF的隐表示,利用深度神经网络在人体生理参数与HRTF的隐表示之间建立非线性映射,得到用户有关的个性化模型。为了尽可能恢复个性化HRTF细节信息,对上述模型降维过程中的残差进行线性建模,得到残差模型。对于目标用户,任意空间位置处的个性化的HRTF可以通过集成三个层次下的模型获得,用于生成三维空间音频。最终,实验结果表明,提出的算法可以有效降低HRTF频谱损失,提升对个性化HRTF的预测性能。展开更多
采用主成分分析方法提取头相关传输函数(head-ralated transfer function,HRTF)的个性化系数,计算了影响HRTF的人体参数的拉普拉斯得分,并联合Pearson相关系数提取出对HRTF影响显著的关键人体参数;构建了径向基函数(radial basis functi...采用主成分分析方法提取头相关传输函数(head-ralated transfer function,HRTF)的个性化系数,计算了影响HRTF的人体参数的拉普拉斯得分,并联合Pearson相关系数提取出对HRTF影响显著的关键人体参数;构建了径向基函数(radial basis function,RBF)神经网络,学习关键人体参数到头相关传输函数个性化系数的非线性映射模型,利用简单的人体参数测量估计出待测者的个性化头相关传输函数.通过实验仿真与偏最小二乘回归(partial least squares regression,PLSR)法比较可知。展开更多
与头相关传递函数(Head—related Transfer Functions:HRTFs)的准确、有效建模对于空间听觉的分析研究以及虚拟听觉空间的生成起着关键的作用。本文通过应用新型的面向多目标参数优化的遗传算法(Genetic Algorithm:GA)进行了 HRTFs共用...与头相关传递函数(Head—related Transfer Functions:HRTFs)的准确、有效建模对于空间听觉的分析研究以及虚拟听觉空间的生成起着关键的作用。本文通过应用新型的面向多目标参数优化的遗传算法(Genetic Algorithm:GA)进行了 HRTFs共用声学极点的极零点模型(Common-Acoustical—Pole and Zero:CAPZ)逼近。实验结果表明,GA较改进的Prony设计方法获得了更优的效果。展开更多
基金Supported by the National Natural Science Foundation of China (Grant No. 10374031)
文摘Based on the measurements from 52 Chinese subjects (26 males and 26 females), a high-spatial-resolution head-related transfer function (HRTF) database with corre- sponding anthropometric parameters is established. By using the database, cues relating to sound source localization, including interaural time difference (ITD), interaural level difference (ILD), and spectral features introduced by pinna, are analyzed. Moreover, the statistical relationship between ITD and anthropometric parameters is estimated. It is proved that the mean values of maximum ITD for male and female are significantly different, so are those for Chinese and western sub- jects. The difference in ITD is due to the difference in individual anthropometric parameters. It is further proved that the spectral features introduced by pinna strongly depend on individual; while at high frequencies (f≥ 5.5 kHz), HRTFs are left-right asymmetric. This work is instructive and helpful for the research on bin- aural hearing and applications on virtual auditory in future.
文摘个性化的头相关传输函数(head-related transfer function,HRTF)可以有效改善空间音频质量。针对个性化HRTF难以精确获得的问题,提出了一种基于层级集成的个性化空间音频生成方法。该方法通过三个模型逐层建立个性化HRTF中的定位信息。首先,采用高斯混合模型建立用户无关的共用模型。然后,采用自编码器获得与用户有关的HRTF的隐表示,利用深度神经网络在人体生理参数与HRTF的隐表示之间建立非线性映射,得到用户有关的个性化模型。为了尽可能恢复个性化HRTF细节信息,对上述模型降维过程中的残差进行线性建模,得到残差模型。对于目标用户,任意空间位置处的个性化的HRTF可以通过集成三个层次下的模型获得,用于生成三维空间音频。最终,实验结果表明,提出的算法可以有效降低HRTF频谱损失,提升对个性化HRTF的预测性能。
文摘采用主成分分析方法提取头相关传输函数(head-ralated transfer function,HRTF)的个性化系数,计算了影响HRTF的人体参数的拉普拉斯得分,并联合Pearson相关系数提取出对HRTF影响显著的关键人体参数;构建了径向基函数(radial basis function,RBF)神经网络,学习关键人体参数到头相关传输函数个性化系数的非线性映射模型,利用简单的人体参数测量估计出待测者的个性化头相关传输函数.通过实验仿真与偏最小二乘回归(partial least squares regression,PLSR)法比较可知。
文摘与头相关传递函数(Head—related Transfer Functions:HRTFs)的准确、有效建模对于空间听觉的分析研究以及虚拟听觉空间的生成起着关键的作用。本文通过应用新型的面向多目标参数优化的遗传算法(Genetic Algorithm:GA)进行了 HRTFs共用声学极点的极零点模型(Common-Acoustical—Pole and Zero:CAPZ)逼近。实验结果表明,GA较改进的Prony设计方法获得了更优的效果。
