Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition

Conformer-based models have become the most dominant end- to-end architecture for speech processing tasks. In this work, we propose a carefully redesigned Conformer with a new down- sampling schema. The proposed model, named Fast Con- former, is 2.8x faster than original Conformer, while preserv- ing state-of-the-art accuracy on Automatic Speech Recognition benchmarks. Also we replace the original Conformer global attention with limited context attention post-training to enable transcription of an hour-long audio. We further improve long- form speech transcription by adding a global token. Fast Con- former combined with a Transformer decoder also outperforms the original Conformer in accuracy and in speed for Speech Translation and Spoken Language Understanding.