Rethinking CNN Models for Audio Classification

TL;DR

This paper demonstrates that ImageNet-pretrained CNNs are effective for audio classification using spectrograms, achieving state-of-the-art results and highlighting the benefits of transfer learning and ensemble methods.

Contribution

It systematically studies the transferability of ImageNet-pretrained CNNs to audio spectrograms and shows their effectiveness for audio classification tasks.

Findings

Pretrained CNNs outperform randomly initialized models.

Visualization reveals what CNNs learn from spectrograms.

Ensemble models improve accuracy significantly.

Abstract

In this paper, we show that ImageNet-Pretrained standard deep CNN models can be used as strong baseline networks for audio classification. Even though there is a significant difference between audio Spectrogram and standard ImageNet image samples, transfer learning assumptions still hold firmly. To understand what enables the ImageNet pretrained models to learn useful audio representations, we systematically study how much of pretrained weights is useful for learning spectrograms. We show (1) that for a given standard model using pretrained weights is better than using randomly initialized weights (2) qualitative results of what the CNNs learn from the spectrograms by visualizing the gradients. Besides, we show that even though we use the pretrained model weights for initialization, there is variance in performance in various output runs of the same model. This variance in performance is due to the random initialization of linear classification layer and random mini-batch orderings in multiple runs. This brings significant diversity to build stronger ensemble models with an overall improvement in accuracy. An ensemble of ImageNet pretrained DenseNet achieves 92.89% validation accuracy on the ESC-50 dataset and 87.42% validation accuracy on the UrbanSound8K dataset which is the current state-of-the-art on both of these datasets.