[Paper Review] Improved Method for Individualization of Head-Related Transfer Functions on Horizontal Plane Using Reduced Number of Anthropometric Measurements

Wahidin Wahab Hugeng, Dadang Gunawan|arXiv (Cornell University)|May 27, 2010

Hearing Loss and Rehabilitation6 references17 citations

TL;DR

This paper proposes an improved method for individualizing head-related transfer functions (HRTFs) on the horizontal plane using only eight key anthropometric measurements. By applying principal component analysis (PCA) to model HRTF magnitude responses and multiple linear regression (MLR) to link PCA weights to these reduced measurements, the method achieves accurate HRTF individualization with minimal user data, enabling natural-sounding spatial audio perception.

ABSTRACT

An important problem to be solved in modeling head-related impulse responses (HRIRs) is how to individualize HRIRs so that they are suitable for a listener. We modeled the entire magnitude head-related transfer functions (HRTFs), in frequency domain, for sound sources on horizontal plane of 37 subjects using principal components analysis (PCA). The individual magnitude HRTFs could be modeled adequately well by a linear combination of only ten orthonormal basis functions. The goal of this research was to establish multiple linear regression (MLR) between weights of basis functions obtained from PCA and fewer anthropometric measurements in order to individualize a given listener's HRTFs with his or her own anthropomety. We proposed here an improved individualization method based on MLR of weights of basis functions by utilizing 8 chosen out of 27 anthropometric measurements. Our objective experiments' results show a superior performance than that of our previous work on individualizing minimum phase HRIRs and also better than similar research. The proposed individualization method shows that the individualized magnitude HRTFs could approximated well the the original ones with small error. Moving sound employing the reconstructed HRIRs could be perceived as if it was moving around the horizontal plane.

Motivation & Objective

• To address the challenge of personalizing HRTFs for individual listeners to improve spatial audio perception.
• To reduce the number of required anthropometric measurements without sacrificing HRTF reconstruction accuracy.
• To develop a regression model linking PCA-based HRTF basis function weights to simplified physical measurements.
• To enable efficient and accurate individualization of HRTFs using minimal listener-specific data.
• To validate the method's superiority over prior approaches in reconstructing original HRTFs with low error.

Proposed method

• Principal component analysis (PCA) was applied to magnitude HRTFs of 37 subjects to extract ten orthonormal basis functions representing the HRTF variation on the horizontal plane.
• The weights of these basis functions were computed for each subject to represent their individual HRTF characteristics.
• Multiple linear regression (MLR) was used to model the relationship between the PCA weights and 27 anthropometric measurements.
• Eight key anthropometric measurements were selected from the 27 through feature selection to minimize redundancy and maximize predictive power.
• The resulting MLR model predicts individual HRTF weights from only eight physical measurements, enabling personalized HRTF synthesis.
• The reconstructed HRTFs were evaluated by comparing them to original HRTFs using error metrics and subjective listening tests for moving sound localization.

Experimental results

Research questions

• RQ1 Can HRTFs be accurately individualized using only a reduced set of anthropometric measurements?
• RQ2 How does the performance of the proposed MLR-based method compare to previous individualization techniques?
• RQ3 To what extent do the selected eight anthropometric measurements capture the variability in HRTF responses across individuals?
• RQ4 Can the reconstructed HRTFs support natural perception of moving sound sources in the horizontal plane?
• RQ5 Is the PCA-MLR framework more efficient and accurate than prior methods using more measurements?

Key findings

• The proposed method achieved superior performance in HRTF individualization compared to the authors' previous work on minimum-phase HRIRs.
• The method outperformed similar existing research in terms of reconstruction accuracy and perceptual quality.
• Using only eight anthropometric measurements, the individualized HRTFs closely approximated the original HRTFs with minimal error.
• Subjects perceived moving sound sources as if they were rotating around the head when using the reconstructed HRIRs.
• The PCA-based model required only ten orthonormal basis functions to represent the entire magnitude HRTF variation across subjects.
• The selected eight measurements were sufficient to predict HRTF weights with high fidelity, reducing data collection burden without compromising accuracy.