ATGNN: Audio Tagging Graph Neural Network
This work addresses audio tagging for sound classification, offering a novel method that is incremental in improving flexibility and efficiency over existing models.
The authors tackled the problem of limited receptive fields in CNNs and inflexible patch processing in Transformers for audio tagging by proposing ATGNN, a graph neural network that treats spectrograms as graphs, achieving 0.585 mAP on FSD50K and 0.335 mAP on AudioSet-balanced with fewer parameters than Transformers.
Deep learning models such as CNNs and Transformers have achieved impressive performance for end-to-end audio tagging. Recent works have shown that despite stacking multiple layers, the receptive field of CNNs remains severely limited. Transformers on the other hand are able to map global context through self-attention, but treat the spectrogram as a sequence of patches which is not flexible enough to capture irregular audio objects. In this work, we treat the spectrogram in a more flexible way by considering it as graph structure and process it with a novel graph neural architecture called ATGNN. ATGNN not only combines the capability of CNNs with the global information sharing ability of Graph Neural Networks, but also maps semantic relationships between learnable class embeddings and corresponding spectrogram regions. We evaluate ATGNN on two audio tagging tasks, where it achieves 0.585 mAP on the FSD50K dataset and 0.335 mAP on the AudioSet-balanced dataset, achieving comparable results to Transformer based models with significantly lower number of learnable parameters.