“Conformer: Convolution-Augmented Transformer for Speech Recognition”, Anmol Gulati, James Qin, Chung-Cheng Chiu, Niki Parmar, Yu Zhang, Jiahui Yu, Wei Han, Shibo Wang, Zhengdong Zhang, Yonghui Wu, Ruoming Pang2020-05-16 (multi-scale Transformers; similar)⁠:

[cf. “Universal Paralinguistic Speech Representations Using Self-Supervised Conformers”, Shor et al 2021] Recently, Transformer and Convolution neural network (CNN) based models have shown promising results in Automatic Speech Recognition (ASR), outperforming Recurrent Neural Networks (RNNs). Transformer models are good at capturing content-based global interactions, while CNNs exploit local features effectively.

In this work, we achieve the best of both worlds by studying how to combine convolution neural networks and transformers to model both local and global dependencies of an audio sequence in a parameter-efficient way. To this regard, we propose the convolution-augmented transformer for speech recognition, named Conformer.

Conformer substantially outperforms the previous Transformer and CNN based models achieving state-of-the-art accuracies. On the widely used LibriSpeech benchmark, our model achieves a Word Error Rate (WER) of 2.1%/4.3% without using a language model and 1.9%/3.9% with an external language model on test/test_other. We also observe competitive performance of 2.7%/6.3% with a small model of only 10M parameters.

Similar Links: