Michele Geronazzo

Enhancing vertical localization with image-guided selection of non-individual head-related transfer functions

A novel approach to the selection of generic head-related transfer functions (HRTFs) for binaural audio rendering through headphones is formalized and described in this paper. A reflection model applied to the users ear picture facilitates extraction of the relevant anthropometric cues that are used for selecting two HRTF sets in a database fitting that user, whose localization performances are evaluated in a complete psychoacoustic experiment. The proposed selection increases the average elevation performances of 17% (with a peak of 34%) with respect to generic HRTFs from an anthropomorphic mannequin. It also significantly enhances externalization and reduces the number of up/down reversals.

A Head-Related Transfer Function Model for Real-Time Customized 3-D Sound Rendering

This paper addresses the problem of modeling head-related transfer functions (HRTFs) for 3-D audio rendering in the front hemisphere. Following a structural approach, we build a model for real-time HRTF synthesis which allows to control separately the evolution of different acoustic phenomena such as head diffraction, ear resonances, and reflections through the design of distinct filter blocks. Parameters to be fed to the model are both derived from mean spectral features in a collection of measured HRTFs and anthropometric features of the specific subject (taken from a photograph of his/her outer ear), hence allowing model customization. Visual analysis of the synthesized HRTFs reveals a convincing correspondence between original and reconstructed spectral features in the chosen spatial range. Furthermore, a possible experimental setup for dynamic psycho acoustical evaluation of such model is depicted.

Fitting pinna-related transfer functions to anthropometry for binaural sound rendering

This paper faces the general problem of modeling pinna-related transfer functions (PRTFs) for 3-D sound rendering. Following a structural approach, we aim at constructing a model for PRTF synthesis which allows to control separately the evolution of ear resonances and spectral notches through the design of two distinct filter blocks. Taking such model as endpoint, we propose a method based on the McAulay-Quatieri partial tracking algorithm to extract the frequencies of the most important spectral notches. Ray-tracing analysis performed on the so obtained tracks reveals a convincing correspondence between extracted frequencies and pinna geometry of a bunch of subjects.

Mixed structural modeling of head-related transfer functions for customized binaural audio delivery

A novel approach to the modeling of head-related transfer functions (HRTFs) for binaural audio rendering is formalized and described in this paper. Mixed structural modeling (MSM) can be seen as the generalization and extension of the structural modeling approach first defined by Brown and Duda back in 1998. Possible solutions for building partial HRTFs (pHRTFs) of the head, torso, and pinna of a specific listener are first described and then used in the construction of two possible mixed structural models of a KEMAR mannequin. Thanks to the flexibility of the MSM approach, an exponential number of solutions for building custom binaural audio displays can be considered and evaluated, the final aim of the process being the achievement of a HRTF model fully customizable by the listener.

Synthetic individual binaural audio delivery by pinna image processing

Purpose – The purpose of this paper is to present a system for customized binaural audio delivery based on the extraction of relevant features from a 2-D representation of the listener’s pinna. Design/methodology/approach – The most significant pinna contours are extracted by means of multi-flash imaging, and they provide values for the parameters of a structural head-related transfer function (HRTF) model. The HRTF model spatializes a given sound file according to the listener’s head orientation, tracked by sensor-equipped headphones, with respect to the virtual sound source. Findings – A preliminary localization test shows that the model is able to statically render the elevation of a virtual sound source better than non-individual HRTFs. Research limitations/implications – Results encourage a deeper analysis of the psychoacoustic impact that the individualized HRTF model has on perceived elevation of virtual sound sources. Practical implications – The model has low complexity and is suitable for implem...

Influence of voxelization on finite difference time domain simulations of head-related transfer functions

The scattering around the human pinna that is captured by the Head-Related Transfer Functions (HRTFs) is a complex problem that creates uncertainties in both acoustical measurements and simulations. Within the simulation framework of Finite Difference Time Domain (FDTD) with axis-aligned staircase boundaries resulting from a voxelization process, the voxelization-based uncertainty propagating in the HRTF-captured sound field is quantified for one solid and two surface voxelization algorithms. Simulated results utilizing a laser-scanned mesh of Knowles Electronics Manikin for Acoustic Research (KEMAR) show that in the context of complex geometries with local topology comparable to grid spacing such as the human pinna, the voxelization-related uncertainties in simulations emerge at lower frequencies than the generally used accuracy bandwidths. Numerical simulations show that the voxelization process induces both random error and algorithm-dependent bias in the simulated HRTF spectral features. Frequencies fr below which the random error is bounded by various dB thresholds are estimated and predicted. Particular shortcomings of the used voxelization algorithms are identified and the influence of the surface impedance on the induced errors is studied. Simulations are also validated against measurements.

Automatic extraction of pinna edges for binaural audio customization

The contribution of the external ear to the head-related transfer function (HRTF) heavily depends on the listeners unique anthropometry. In particular, the shape of the most prominent contours of the pinna defines the frequency location of the HRTF spectral notches along the elevation of the sound source. This paper addresses the issue of automatically estimating the location of pinna edges starting from a set of pictures produced by a multi-flash imaging device. A basic image processing algorithm designed to obtain the principal edges and their distance from the ear canal entrance is described. The effectiveness of the developed hardware and software is preliminarily evaluated on a small number of test subjects.

Acoustic selfies for extraction of external ear features in mobile audio augmented reality

Virtual and augmented realities are expected to become more and more important in everyday life in the next future; the role of spatial audio technologies over headphones will be pivotal for application scenarios which involve mobility. This paper introduces the SelfEar project, aimed at low-cost acquisition and personalization of Head-Related Transfer Functions (HRTFs) on mobile devices. This first version focuses on capturing individual spectral features which characterize external ear acoustics, through a self-adjustable procedure which guides users in collecting such information: their mobile device must be held with the stretched arm and positioned at several specific elevation points; acoustic data are acquired by an audio augmented reality headset which embeds a pair of microphones at listener ear-canals. A preliminary measurement session assesses the ability of the system to capture spectral features which are crucial for elevation perception. Moreover, a virtual experiment using a computational auditory model predicts clear vertical localization cues in the measured features.

Localization of self-generated synthetic footstep sounds on different walked-upon materials through headphones

AbstractThis paper focuses on the localization of footstep sounds interactively generated during walking and provided through headphones. Three distinct experiments were conducted in a laboratory involving a pair of sandals enhanced with pressure sensors and a footstep synthesizer capable of simulating two typologies of surface materials: solid (e.g., wood) and aggregate (e.g., gravel). Different sound delivery methods (mono, stereo, binaural) as well as several surface materials, in the presence or absence of concurrent contextual auditory information provided as soundscapes, were evaluated in a vertical localization task. Results showed that solid surfaces were localized significantly farther from the walker’s feet than the aggregate ones. This effect was independent of the used rendering technique, of the presence of soundscapes, and of merely temporal or spectral attributes of sound. The effect is hypothesized to be due to a semantic conflict between auditory and haptic information such that the higher the semantic incongruence the greater the distance of the perceived sound source from the feet. The presented results contribute to the development of further knowledge toward a basis for the design of continuous multimodal feedback in virtual reality applications .

Extraction of Pinna Features for Customized Binaural Audio Delivery on Mobile Devices

The paper presents a system for customized binaural audio delivery based on the extraction of the relevant features from a 2-D representation of the listeners pinna. A procedure based on multi-flash imaging for recognizing the main contours of the pinna and their position with respect to the ear canal entrance is detailed. The resulting contours drive the parametrization of a structural head-related transfer function model that performs in real time the spatialization of a desired sound file according to the listeners position with respect to the virtual sound source, tracked by sensor-equipped headphones. The low complexity of the model allows smooth implementation and delivery on any mobile device. The purpose of the desired system is to provide low-tech custom binaural audio to any user without the need of tedious and cumbersome subjective measurements.