Estimation of HRTFs on the horizontal plane using physical features

• Published in: Applied acoustics Vol. 68; no. 8; pp. 897 - 908
• Main Authors: Nishino, Takanori, Inoue, Naoya, Takeda, Kazuya, Itakura, Fumitada
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01.08.2007; Elsevier
• Subjects: Acoustical measurements and instrumentation; Acoustics; Exact sciences and technology; Fundamental areas of phenomenology (including applications); Head related transfer function; Multiple regression analysis; Physical feature; Physics; Principal component analysis; Sound localization; 43.66.Pn; 43.66.Dc; 43.66.Qp; Multiple regression analysis; Sound localization; Physical feature; Head related transfer function; Principal component analysis; Distortion; Acoustics; Sound; Regression analysis; Physical properties; Multiple regression; Head related transfer function; Physical feature; Multiple regression analysis; Principal component analysis; Sound localization; 43.66.Dc; 43.66.Pn; 43.66.Qp; Bandwidth; Experimentation; Measurement method; Headphone; Transfer function; Localization; Performance
• ISSN: 0003-682X; 1872-910X
• DOI: 10.1016/j.apacoust.2006.12.010

Sound localization can be controlled by using head related transfer functions (HRTFs), which are related to the size of the head, the ears and so on. Since HRTFs are characterized by source directions and subjects, it is necessary to conduct measurements in all directions for all subjects. However, such measurement is expensive and time-consuming. In this paper, we propose a simpler and more useful method that investigates the relationship between HRTFs and physical size by multiple regression analysis. The estimated HRTFs are evaluated by objective and subjective measures. For objective results, the average spectral distortion score is 4.0 dB in a bandwidth ranging from 0 to 8 kHz. Subjective results indicate no significant difference between the measured and the estimated HRTFs in that frequency range. These results support the hypothesis that the proposed method is effective for estimating HRTFs.