XAI-based Comparison of Input Representations for Audio Event Classification
This work addresses the problem of selecting optimal audio input representations for researchers and practitioners in audio event classification, though it is incremental as it applies existing XAI methods to compare known representations.
The study used explainable AI to compare how deep neural networks trained on raw waveforms versus spectrograms classify audio events, revealing representation-dependent decision strategies and enabling a well-informed choice for robust and representative input representations.
Deep neural networks are a promising tool for Audio Event Classification. In contrast to other data like natural images, there are many sensible and non-obvious representations for audio data, which could serve as input to these models. Due to their black-box nature, the effect of different input representations has so far mostly been investigated by measuring classification performance. In this work, we leverage eXplainable AI (XAI), to understand the underlying classification strategies of models trained on different input representations. Specifically, we compare two model architectures with regard to relevant input features used for Audio Event Detection: one directly processes the signal as the raw waveform, and the other takes in its time-frequency spectrogram representation. We show how relevance heatmaps obtained via "Siren"{Layer-wise Relevance Propagation} uncover representation-dependent decision strategies. With these insights, we can make a well-informed decision about the best input representation in terms of robustness and representativity and confirm that the model's classification strategies align with human requirements.