Skip to Main content Skip to Navigation

Individualisation d'indices acoustiques pour la synthèse binaurale

Abstract : Binaural synthesis is a sound spatialization technology, which is the closest to na-
tural hearing. It allows the spatialization of a monophonic sound source with only two
filters for a given position. The filters are defined by the HRTFs (Head Related Transfer
Function) corresponding to the left and right ear of the listener. The major drawback of
binaural synthesis is that the HRTF, which are related to the listener's morphology, are
strongly individual. Listening with non-individual HRTF could lead to audible artifacts.
Therefore binaural rendering of high quality requires individualized HRTF. This thesis
tackles the problem of the individualization of binaural synthesis in the framework of
its implementation as a pure delay, the interaural time di®erence (ITD), and a minimal
phase filter determined by the magnitude of the HRTF. The work conducted on the ITD
validates the implementation chosen even for the positions where the HRTF are poorly
minimum phase filters. In addition the ITD calculation methods which are close to per-
ception are pointed out. An experimental study is also undertaken to investigate the
resolution of the ITD with the elevation angle along the cones of confusion. Perceptual
results indicate that the ITD variation with the elevation angle needs to be reproduced.
In order to account for this variation, a new formula is proposed on the basis of the
spherical head model. Optimization of the parameters of this formula for a whole ITD
database provides an average formulation which is appropriate for a large number of sub-
jects and for many applications. Concerning the modeling of the spectral cues (HRTF
magnitude), the Boundary Element Method (BEM) has been examined. It is concluded
that BEM methods are useful in combination with measurement for the modeling of
the low frequency part. A new approach, which involves statistical learning technique, is
proposed for the HRTF prediction. A neural network is built to compute HRTF in any
direction from a limited set of measured HRTF. Preliminary assessment of this modeling
shows that the neural network succeeds well in individualizing spectral cues. This result
suggests a simplified protocol of HRTF measurement : HRTF are measured for only a
few directions and the HRTF for the other locations are obtained by the neural network.
Complete list of metadatas
Contributor : Sylvain Busson <>
Submitted on : Friday, August 4, 2006 - 4:10:27 AM
Last modification on : Tuesday, October 20, 2020 - 3:10:21 AM
Long-term archiving on: : Friday, November 25, 2016 - 11:22:35 AM


  • HAL Id : tel-00012023, version 2


Sylvain Busson. Individualisation d'indices acoustiques pour la synthèse binaurale. Acoustique [physics.class-ph]. Université de la Méditerranée - Aix-Marseille II, 2006. Français. ⟨tel-00012023v2⟩



Record views


Files downloads