Abstract:
Nowadays, the “state of the art” in the binaural spatialization field is rather advanced: on the market and, most of all, in various research centres all over the world, it is possible to find different kinds of algorithms for the binaural simulation of virtual sound sources in a three-dimensional soundscape. Nevertheless, the simulation performed by these algorithms is mainly related to the azimuth and the elevation parameters; only few of them implement a distance simulation technique.
In this paper a technique will be presented for the binaural simulation of distance. Studying the mechanisms used by our hearing system for the estimation of the distance of a sound source, three “distance cues” have been identified and simulated individually:
• The attenuation of the air, which can be approximated as linear in frequency response until about 15 meters of distance. This cue has been simulated through a simple variable reduction and filtering line.
• The differences in the HRTF spectrum, mainly related to close sound sources (between 0 and 3 meters). This cue has been simulated through the convolution with 5 different HRTFs extracted at different distances.
• The direct to reflected sound signal ratio (of course, this is not relevant in a free-field or anechoic space). This cue has been simulated splitting the direct, first reflections and reverberant binaural impulse parts, and mixing, with different weightings, the differently spatialized signals.
The signal in input to the algorithm is therefore filtered and spatialized using cross-fades and weighted sums between signals convolved with different HRTFs: the initial processing results have been supported by the data coming from a preliminary simple subjective testing stage performed with 25 individuals. Further testing is planned.
