SpeechNet: Weakly Supervised, End-to-End Speech Recognition at Industrial Scale
Abstract
End-to-end automatic speech recognition systems represent the state of the art, but they rely on thousands of hours of manually annotated speech for training, as well as heavyweight computation for inference. Of course, this impedes commercialization since most companies lack vast human and computational resources. In this paper, we explore training and deploying an ASR system in the label-scarce, compute-limited setting. To reduce human labor, we use a third-party ASR system as a weak supervision source, supplemented with labeling functions derived from implicit user feedback. To accelerate inference, we propose to route production-time queries across a pool of CUDA graphs of varying input lengths, the distribution of which best matches the traffic's. Compared to our third-party ASR, we achieve a relative improvement in word-error rate of 8% and a speedup of 600%. Our system, called SpeechNet, currently serves 12 million queries per day on our voice-enabled smart television. To our knowledge, this is the first time a large-scale, Wav2vec-based deployment has been described in the academic literature.
Cite
@article{arxiv.2211.11740,
title = {SpeechNet: Weakly Supervised, End-to-End Speech Recognition at Industrial Scale},
author = {Raphael Tang and Karun Kumar and Gefei Yang and Akshat Pandey and Yajie Mao and Vladislav Belyaev and Madhuri Emmadi and Craig Murray and Ferhan Ture and Jimmy Lin},
journal= {arXiv preprint arXiv:2211.11740},
year = {2022}
}
Comments
Accepted to EMNLP 2022 Industry Track; 9 pages, 7 figures