Binaural Source Localization based on Modulation-Domain Features and Decision Pooling
This work addresses source localization for hearing aids to improve signal-to-noise ratio, but it is incremental as it builds on existing methods with specific feature improvements.
The paper tackled the problem of binaural source localization by applying Amplitude Modulation Spectrum features and a binaurally trained classifier, achieving a 4.25° smaller mean absolute error than the baseline on the LOCATA dataset.
In this work we apply Amplitude Modulation Spectrum (AMS) features to the source localization problem. Our approach computes 36 bilateral features for 2s long signal segments and estimates the azimuthal directions of a sound source through a binaurally trained classifier. This directional information of a sound source could be e.g. used to steer the beamformer in a hearing aid to the source of interest in order to increase the SNR. We evaluated our approach on the development set of the IEEE-AASP Challenge on sound source localization and tracking (LOCATA) and achieved a 4.25° smaller MAE than the baseline approach. Additionally, our approach is computationally less complex.