Master 2013 2014
Stages de la spécialité SAR
Binaural Listening : grouping of Head-Related Transfer Function and non-perceptual reciprocity

Site :Audio & Acoustique
Lieu :LIMSI-CNRS, Orsay
Encadrant : &
Dates :period normal
Rémunération :suivant la grille standard du CNRS
Mots-clés : Parcours ATIAM : Acoustique, Parcours ATIAM : Traitement du signal, Parcours SAR autre qu’ATIAM, recherche


Binaural sound reproduction attempts to reproduce at a listener’s ears the acoustic signals that would have arrived at his ears if he had been at the original sound scene. This technique can produce sound images anywhere around the listener in a 3D space, based on interaural time and level differences and monoral spectral cues.

However, the filters that describe interaural differences and monoral cues as a function of a sound source’s position, the Head-Related Transfer Functions (HRTF), are individual. They are related to each individual’s morphology. Listening to binaural recording that have been designed using another individual’s HRTFs results in incorrect localization.

The aim of this internship is twofold : • To improve a protocol that has been experimented in the past, the student will have to evaluate how well different sets of HRTFs can describe a theoretical sound source’s trajectory. • The previous study showed that there is no systematic reciprocity of the ratings of non-individual sets of HRTFs : if subject A rates very good the HRTFs of subject B, subject B might not rate subject A’s HRTFs very good too. The intern will try to find which parts of HRTFs show reciprocity and which don’t.

Highly motivated and dynamic candidates are encouraged to apply. No a priori knowledge of spatial audio is required, though interest and willingness to learn are beneficial.

Profile : Master’s level M2 in acoustics, psychoacoustics, spatial audio, or related fields. Autonomy and an inquisitive mind are important.

Required skills : Matlab programming and good English for the bibliographic review are mandatory. Knowledge of statistical analysis, 3D audio will be appreciated.