Two aspects of stereo imaging accuracy from audio system listening have been investigated: (i) panned phantom image localization accuracy at 5-degree and (ii) sweet spot spatial spread from the ideal anechoic reference. The simulated study used loudspeakers of different directivity under ideal anechoic or realistic varying reverberant room conditions and extracted binaural auditory features (ILDs, ITDs, and ICs) from the received audio signals. For evaluation, a Decision Tree classifier was used under a sparse data self-training achieving localization accuracy ranging from 92% (for ideal anechoic when training/test data were similar audio category), down to 55% (for high reverberation when training/test data were different music segments).Sweet spot accuracy was defined and evaluated as a spatial spread statistical distribution function.
Kamaris, Gavriil; Karlos, Stamatis; Fazakis, Nikos; Terpinas, Stergios; Mourjopoulos, John
Affiliation: University of Patras, Patras, Greece
AES Convention:140 (May 2016) eBrief:269
Publication Date:May 26, 2016
No AES members have commented on this paper yet.
If you are not yet an AES member and have something important to say about this paper then we urge you to join the AES today and make your voice heard. You can join online today by clicking here.