CAT: Causal Audio Transformer for Audio Classification

Xiaoyu Liu; Hanlin Lu; Jianbo Yuan; Xinyu Li

ICASSP 2023accepted0 citations

CAT: Causal Audio Transformer for Audio Classification

Xiaoyu Liu, Hanlin Lu, Jianbo Yuan, Xinyu Li

Abstract

The attention-based Transformers have been increasingly applied to audio classification because of their global receptive field and ability to handle long-term dependency. However, the existing frameworks which are mainly extended from the Vision Transformers are not perfectly compatible with audio signals. In this paper, we introduce a Causal Audio Transformer (CAT) consisting of a Multi-Resolution Multi- Feature (MRMF) feature extraction with an acoustic attention block for more optimized audio modeling. In addition, we propose a causal module that alleviates over-fitting, helps with knowledge transfer and improves interpretability. CAT obtains higher or comparable state-of-the-art classification performance on ESC50, AudioSet and UrbanSound8K datasets, and can be easily generalized to other Transformer- based models.

BibTeX

@inproceedings{icassp2023_catcausalaudiotr,
  title = {CAT: Causal Audio Transformer for Audio Classification},
  author = {Xiaoyu Liu and Hanlin Lu and Jianbo Yuan and Xinyu Li},
  booktitle = {ICASSP 2023},
  year = {2023}
}