English

Developing Real-time Streaming Transformer Transducer for Speech Recognition on Large-scale Dataset

Computation and Language 2021-03-02 v3 Audio and Speech Processing

Abstract

Recently, Transformer based end-to-end models have achieved great success in many areas including speech recognition. However, compared to LSTM models, the heavy computational cost of the Transformer during inference is a key issue to prevent their applications. In this work, we explored the potential of Transformer Transducer (T-T) models for the fist pass decoding with low latency and fast speed on a large-scale dataset. We combine the idea of Transformer-XL and chunk-wise streaming processing to design a streamable Transformer Transducer model. We demonstrate that T-T outperforms the hybrid model, RNN Transducer (RNN-T), and streamable Transformer attention-based encoder-decoder model in the streaming scenario. Furthermore, the runtime cost and latency can be optimized with a relatively small look-ahead.

Keywords

Cite

@article{arxiv.2010.11395,
  title  = {Developing Real-time Streaming Transformer Transducer for Speech Recognition on Large-scale Dataset},
  author = {Xie Chen and Yu Wu and Zhenghao Wang and Shujie Liu and Jinyu Li},
  journal= {arXiv preprint arXiv:2010.11395},
  year   = {2021}
}

Comments

5 pages

R2 v1 2026-06-23T19:32:25.428Z