AND: Audio Network Dissection for Interpreting Deep Acoustic Models

Neuron-level interpretations aim to explain network behaviors and properties by investigating neurons responsive to specific perceptual or structural input patterns. Although there is emerging work in the vision and language domains, none is explored for acoustic models. To bridge the gap, we introduce textitAND, the first Audio Network Dissection framework that automatically establishes natural language explanations of acoustic neurons based on highly-responsive audio. textitAND features the use of LLMs to summarize mutual acoustic features and identities among audio. Extensive experiments are conducted to verify textitAND's precise and informative descriptions. In addition, we demonstrate a potential use of textitAND for audio machine unlearning by conducting concept-specific pruning based on the generated descriptions. Finally, we highlight two acoustic model behaviors with analysis by textitAND: (i) models discriminate audio with a combination of basic acoustic features rather than high-level abstract concepts; (ii) training strategies affect model behaviors and neuron interpretability -- supervised training guides neurons to gradually narrow their attention, while self-supervised learning encourages neurons to be polysemantic for exploring high-level features.