We design an online end-to-end speech recognition system based on Time-Depth Separable (TDS) convolutions and Connectionist Temporal Classification (CTC). We improve the core TDS architecture in order to limit the future context and hence reduce latency while maintaining accuracy. The system has almost three times the throughput of a well tuned hybrid ASR baseline while also having lower latency and a better word error rate. Also important to the efficiency of the recognizer is our highly optimized beam search decoder. To show the impact of our design choices, we analyze throughput, latency, accuracy, and discuss how these metrics can be tuned based on the user requirements.
Cite as: Pratap, V., Xu, Q., Kahn, J., Avidov, G., Likhomanenko, T., Hannun, A., Liptchinsky, V., Synnaeve, G., Collobert, R. (2020) Scaling Up Online Speech Recognition Using ConvNets. Proc. Interspeech 2020, 3376-3380, doi: 10.21437/Interspeech.2020-2840
@inproceedings{pratap20b_interspeech,
author={Vineel Pratap and Qiantong Xu and Jacob Kahn and Gilad Avidov and Tatiana Likhomanenko and Awni Hannun and Vitaliy Liptchinsky and Gabriel Synnaeve and Ronan Collobert},
title={{Scaling Up Online Speech Recognition Using ConvNets}},
year=2020,
booktitle={Proc. Interspeech 2020},
pages={3376--3380},
doi={10.21437/Interspeech.2020-2840}
}