Individualized HRTFs From Few Measurements: a Statistical Learning Approach

Vincent Lemaire (1) , Fabrice Clérot (1) , Sylvain Busson (1, 2) , Rozenn Nicol (1) , Vincent V. Choqueuse (3, 4)

1 FTR&D - France Télécom Recherche et Développement [Lannion]
2 LMA - Laboratoire de Mécanique et d'Acoustique [Marseille]
3 Lab-STICC - Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance (UMR 3192)
4 LBMS - Laboratoire brestois de mécanique et des systèmes

Virtual Auditory Space (VAS) refers to the synthesis and simulation of spatial hearing using earphones and/or a speaker system. High-fidelity VAS requires the use of individualized head-related transfer functions (HRTFs) which describe the acoustic filtering properties of the listener's external auditory periphery. HRTFs serve the increasingly dominant role of implementation 3-D audio systems, which have been realized in some commercial applications. However, the cost of a 3-D audio system cannot be brought down because the efficiency of computation, the size of memory, and the synthesis of unmeasured HRTFs remain to be made better. Because HRTFs are unique for each user depending on his morphology, the economically realist synthesis of individualized HRTFs has to rely on some measurements. This paper presents a way to reduce the cost of a 3-D audio system using a statistical modeling which allows to use only few measurements for each user.

Mots clés: HRTF; Signal Processing; Neural Networks; Kohonen Map

Domaines: Acoustique [physics.class-ph]; Acoustique [physics.class-ph]; Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Vincent Lemaire, Fabrice Clérot, Sylvain Busson, Rozenn Nicol, Vincent V. Choqueuse. Individualized HRTFs From Few Measurements: a Statistical Learning Approach. Proceedings of International Joint Conference on Neural Networks, Jul 2005, Montreal, Canada. pp.2041 - 2046, ⟨10.1109/IJCNN.2005.1556214⟩. ⟨hal-00493958⟩