English

Exploring SSL Discrete Speech Features for Zipformer-based Contextual ASR

Computation and Language 2025-06-11 v2 Sound Audio and Speech Processing

Abstract

Self-supervised learning (SSL) based discrete speech representations are highly compact and domain adaptable. In this paper, SSL discrete speech features extracted from WavLM models are used as additional cross-utterance acoustic context features in Zipformer-Transducer ASR systems. The efficacy of replacing Fbank features with discrete token features for modelling either cross-utterance contexts (from preceding and future segments), or current utterance's internal contexts alone, or both at the same time, are demonstrated thoroughly on the Gigaspeech 1000-hr corpus. The best Zipformer-Transducer system using discrete tokens based cross-utterance context features outperforms the baseline using utterance internal context only with statistically significant word error rate (WER) reductions of 0.32% to 0.41% absolute (2.78% to 3.54% relative) on the dev and test data. The lowest published WER of 11.15% and 11.14% were obtained on the dev and test sets. Our work is open-source and publicly available at https://github.com/open-creator/icefall/tree/master/egs/gigaspeech/Context\_ASR.

Keywords

Cite

@article{arxiv.2409.08797,
  title  = {Exploring SSL Discrete Speech Features for Zipformer-based Contextual ASR},
  author = {Mingyu Cui and Yifan Yang and Jiajun Deng and Jiawen Kang and Shujie Hu and Tianzi Wang and Zhaoqing Li and Shiliang Zhang and Xie Chen and Xunying Liu},
  journal= {arXiv preprint arXiv:2409.08797},
  year   = {2025}
}

Comments

Accepted by Interspeech 2025

R2 v1 2026-06-28T18:43:40.224Z