Conformer: Convolution-augmented Transformer for Speech Recognition
This work addresses speech recognition for applications requiring high accuracy, but it is incremental as it builds on existing transformer and CNN methods.
The paper tackled the problem of combining convolutional neural networks and transformers for speech recognition to model both local and global dependencies efficiently, resulting in state-of-the-art accuracies with a word error rate of 2.1%/4.3% on the LibriSpeech benchmark without a language model.
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 significantly outperforms the previous Transformer and CNN based models achieving state-of-the-art accuracies. On the widely used LibriSpeech benchmark, our model achieves WER of 2.1%/4.3% without using a language model and 1.9%/3.9% with an external language model on test/testother. We also observe competitive performance of 2.7%/6.3% with a small model of only 10M parameters.