Individualization of Head Related Transfer Functions Based on Radial Basis Function Neural Network

Head Related Transfer Functions (HRTFs) contain sound localization cues and are commonly used in 3D audio reproduction. Due to HRTFs are closely related to anthropometric parameters (head, pinna, torso), which means HRTFs vary with each individual, how to obtain a set of suitable HRTFs for each individual remains to be solved. In this paper, we investigated the complex relationship between HRTFs and anthropometric parameters through Radial Basis Function neural network (RBF), and proposed a method of generating individualized HRTFs with listener's anthropometric parameters. Objective experiments show that the estimated HRTFs have good consistency with measured ones, and the spectral distortion values have an average reduction of 0.59 dB compared with other methods. Subjective listening tests show that using estimated HRTFs enable accurate auditory localization.

[1]  Lin Li,et al.  Modeling individual HRTF tensor using high-order partial least squares , 2014, EURASIP J. Adv. Signal Process..

[2]  Zhenyang Wu,et al.  Head-Related Transfer Function Personalization Based on Partial Least Square Regression: Head-Related Transfer Function Personalization Based on Partial Least Square Regression , 2011 .

[3]  Zhenyang Wu,et al.  Head Related Transfer Function Personalization Based on Multiple Regression Analysis , 2006, 2006 International Conference on Computational Intelligence and Security.

[4]  Kazuya Takeda,et al.  Estimation of HRTFs on the horizontal plane using physical features , 2007 .

[5]  B. Shinn-Cunningham,et al.  Tori of confusion: binaural localization cues for sources within reach of a listener. , 2000, The Journal of the Acoustical Society of America.

[6]  M. Alex O. Vasilescu,et al.  A Multilinear (Tensor) Framework for HRTF Analysis and Synthesis , 2007, 2007 IEEE International Conference on Acoustics, Speech and Signal Processing - ICASSP '07.

[7]  C. Avendano,et al.  The CIPIC HRTF database , 2001, Proceedings of the 2001 IEEE Workshop on the Applications of Signal Processing to Audio and Acoustics (Cat. No.01TH8575).

[8]  Xiangyang Zeng,et al.  A hybrid algorithm for selecting head-related transfer function based on similarity of anthropometric structures , 2010 .

[9]  B F Katz,et al.  Boundary element method calculation of individual head-related transfer function. I. Rigid model calculation. , 2001, The Journal of the Acoustical Society of America.

[10]  Larry S. Davis,et al.  HRTF personalization using anthropometric measurements , 2003, 2003 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (IEEE Cat. No.03TH8684).

[11]  Laura E. Ray,et al.  Individualization of head related transfer functions using principal component analysis , 2015 .

[12]  Rajesh M. Hegde,et al.  A joint sparsity and linear regression based method for customization of median plane HRIR , 2015, 2015 49th Asilomar Conference on Signals, Systems and Computers.

[13]  Dadang Gunawan,et al.  The Effectiveness of Chosen Partial Anthropometric Measurements in Individualizing Head-Related Transfer Functions on Median Plane , 2011 .

[14]  Jens Blauert,et al.  The AUDIS catalog of human HRTFs , 1998 .

[15]  F L Wightman,et al.  Localization using nonindividualized head-related transfer functions. , 1993, The Journal of the Acoustical Society of America.

[16]  Dadang Gunawan,et al.  Improved Method for Individualization of Head-Related Transfer Functions on Horizontal Plane Using Reduced Number of Anthropometric Measurements , 2010, ArXiv.