Search

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 twofilters for a given position. The filters are defined by the HRTFs (Head Related TransferFunction) corresponding to the left and right ear of the listener. The major drawback ofbinaural synthesis is that the HRTF, which are related to the listener's morphology, arestrongly individual. Listening with non-individual HRTF could lead to audible artifacts.Therefore binaural rendering of high quality requires individualized HRTF. This thesistackles the problem of the individualization of binaural synthesis in the framework ofits implementation as a pure delay, the interaural time di®erence (ITD), and a minimalphase filter determined by the magnitude of the HRTF. The work conducted on the ITDvalidates the implementation chosen even for the positions where the HRTF are poorlyminimum 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 theresolution of the ITD with the elevation angle along the cones of confusion. Perceptualresults 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 thespherical head model. Optimization of the parameters of this formula for a whole ITDdatabase 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 (HRTFmagnitude), the Boundary Element Method (BEM) has been examined. It is concludedthat BEM methods are useful in combination with measurement for the modeling ofthe low frequency part. A new approach, which involves statistical learning technique, isproposed for the HRTF prediction. A neural network is built to compute HRTF in anydirection from a limited set of measured HRTF. Preliminary assessment of this modelingshows that the neural network succeeds well in individualizing spectral cues. This resultsuggests a simplified protocol of HRTF measurement : HRTF are measured for only afew directions and the HRTF for the other locations are obtained by the neural network., La synthèse binaurale est la technique de spatialisation sonore la plus proche del'écoute naturelle. Elle permet un rendu spatialisé d'une source monophonique à une po-sition donnée avec seulement deux filtres qui correspondent aux oreilles gauche et droite :les HRTF (Head Related Transfer Function). L'inconvénient majeur de la technique bi-naurale repose sur le fait que les HRTF, liées à la morphologie de l'auditeur, sont propresà chaque utilisateur. Une écoute avec des HRTF non-individuelles comporte des artefactsaudibles. Il faut donc acquérir des HRTF individuelles. Cette thèse aborde le problèmede l'individualisation de la synthèse binaurale dans le cadre de son implémentation en unretard pur, la différence interaurale de temps (ITD), et un filtre à phase minimale déter-miné par le module de la HRTF. Le travail sur l'ITD permet de valider l'implémentationchoisie même pour les positions où les HRTF sont mal décrites par des filtres à phaseminimale et permet de déterminer, parmi les méthodes classiques de calcul de l'ITD,celles qui estiment une ITD proche de la perception. Une étude expérimentale est aussimenée pour établir la résolution de l'ITD avec l'angle d'élévation. Les résultats indiquentla nécessité perceptive de reproduire les variations de l'ITD en élévation. Une nouvelleformule d'estimation de l'ITD créée sur la base d'un modèle de tête sphérique, la formulede déplacement des oreilles (FDO), est développée pour rendre compte de ces variations.L'optimisation des paramètres de cette formule aux ITD de toute une base de donnéesde HRTF permet d'entrevoir une formulation moyenne convenant pour un grand nombrede personne et pour de nombreuses applications. L'étude s'est ensuite focalisée sur lamodélisation du module spectral (filtre à phase minimale). Le travail réalisé sur l'appli-cation des méthodes de calcul par éléments de frontière (BEM pour Boundary ElementMethod) pour l'acquisition de HRTF, indique que cette méthode, peut notamment êtreutilisée en complément des mesures pour l'acquisition de la partie basse fréquence desHRTF. Une approche originale, qui applique des techniques d'apprentissage statistique,est proposée et étudiée pour la modélisation de HRTF. Un réseau de neurones artificiels(RNA) est entra^³né pour calculer des HRTF d'un individu à partir de la connaissancedes HRTF mesurées en un nombre réduit de positions. Les premiers résultats sont en-courageants : le modèle permet d'atteindre un degré assez fin d'individualisation, ce quisuggère un protocole simplifié d'acquisition de HRTF. Un faible nombre de mesures estacquis et les autres sont prédites par le modèle.

National audience; Dans cet article, nous prouvons la faisabilité dʼun suivi comportemental des organismes benthiques in situ sans dérangement en écoutant uniquement les sons quʼils produisent. Notre modèle dʼétude est la coquille Saint-Jacques (Pencten maximus). Ces travaux sont menés dʼune part pour identifier et caractériser les sources de bruit ambiant en milieu très côtier et dʼautre part pour définir un outil de suivi du comportement des organismes vivants marins utilisés par la communauté comme archives biologiques. A partir de mesures en laboratoire et de terrain, nous démontrons que lʼactivité de respiration de la coquille Saint-Jacques est associée à une production sonore mesurable. Cette activité, traduite par une ouverture et une fermeture rapide de la coquille, crée un son transitoire large bande à partir de 10 kilohertz dʼune durée de lʼordre dʼune seconde. Le niveau spectral de cette production sonore excède le bruit ambiant en rade de Brest ou dans lʼarchipel de Molène de 8 dB. Grâce à cet excédent, la portée de détection de ces sons est comprise entre 80 centimètres et 120 centimètres.

National audience; The passive acoustic monitoring (PAM) is a tool of choice for non-intrusive study of aquatic organisms in the wild. Anthropogenic disturbance may affect animal behavior and should generally be minimised. In this study, a PAM system was used to record the bioacoustic sound produced by the valve movements of the great scallop (\textit{Pecten maximus}). The primary function of PAM system is to detect bioacoustic emissions embedded in the background noise. Sounds produced by the great scallop are transients. Detecting them with power based detector could lead to misses particularly when signal to noise ratio is weak. In this paper, a click detection scheme based on higher-order statistics is applied to the monitoring of the great scallop in its habitat. The detector is effective in tracking the sound produced by the valve movements of the great scallop. The PAM system in association with the click detector represent a promising tool for monitoring, with minimal anthropogenic disturbance, the behavior of minimally mobile aquatic organisms in the wild.

Passive Acoustic Monitoring (PAM) of marine mammal vocalizations has been intensively used in applications such as marine wildlife surveys. Marine mammals regularly produce sounds to echolocate, to communicate and while foraging and PAM is a complementary tool of visual-based observations for density estimation. Vocalizations of marine mammals are divided into two categories: impulse sounds and frequency modulation whistles. A common way to detect whistles consists in forming signal spectrogram and to check for frequency modulation tracks. Whistles tracks are made of energetic pixels identified by comparing each spectrogram pixel levels with an estimate of local noise. Despite simple to implement and quite effective, usual noise level estimation via signal low-pass filtering lead to false alarms in non-stationary noise conditions as in the vicinity of human activities or in costal observatories. A noise level estimation algorithm is proposed by taking into account the statistical properties of noise only samples of the spectrogram. The algorithm is based on minima statistics of a time-frequency neighborhood of the current spectrogram pixel where noise only pixels follow a $&_chi;^2$ distribution. The estimation is effectively applied on real data. Major drawbacks coming from low-pass filtering are avoided and the noise level is in closer agreement to the real noise level. Techniques are proposed to keep the real-time implementation of the algorithm feasible.

cote interne IRCAM: Busson05b; None / None; National audience; The Interaural Time Difference (ITD) is the primary cue for sound source localization. Although the ITD has been intensively studied, there are still many open questions. This paper focuses on the ITD in relation to minimum phase filter modeling for binaural synthesis. Comparison of the ITD estimate methods highlights discrepancies for locations near the interaural axis. A sub jective experiment was performed to estimate the ITD value to use with a given pair of minimum phase filters. The task of the sub ject consists in adjusting the ITD of the test sound until its perceived location matches the target sound. The psychoacoustic value of the ITD is compared with computational estimates from Head Related Transfer Function (HRTF).

cote interne IRCAM: Busson04a; None / None; National audience; The spatialization rendering with binaural synthesis is still limited by the individualization of the spatial filters, which are usually composed of a pure delay for the interaural time difference (ITD) and a minimumphase filter for the spectral cues. This paper deals with the individualization of the ITD based on physical modeling. Individualized ITDs are well predicted by a spherical-head model with a customized radius derived from an empiric linear combination of three head anthropometric parameters describing the listener’s head. However it can be observed that the constant ITD contours computed from the spherical-head model are concentric circles, whereas the constant ITD contours extracted from measured HRTFs show deviations from these circles, especially for azimuths near to the interaural axis. The location of the ears with respect to the center of the head seems to be the main factor to reproduce these deviations. The authors propose to introduce a parameter governing the ears’ shifting, so as to break the symmetry of the spherical-head model. This idea leads to a new formula for the computation of individualized ITDs. The formula is tested by comparison with measured data. The authors also examine to what extent the deviations of the measured ITD along the circles contribute to the individualization. Psycho-acoustical experiments are performed to answer this question.