Listenable Maps for Audio Classifiers

arXiv:2403.13086v1 Announce Type: cross
Abstract: Despite the impressive performance of deep learning models across diverse tasks, their complexity poses challenges for interpretation. This challenge is particularly evident for audio signals, where conveying interpretations becomes inherently difficult. To address this issue, we introduce Listenable Maps for Audio Classifiers (L-MAC), a posthoc interpretation method that generates faithful and listenable interpretations. L-MAC utilizes a decoder on top of a pretrained classifier to generate binary masks that highlight relevant portions of the input audio. …

abstract arxiv audio challenge challenges classifiers complexity cs.lg cs.sd deep learning diverse eess.as eess.sp interpretation issue mac maps performance tasks type