arXiv：cs.CL· Kiet Anh Hoang, Khanh Le, Bao Nguyen, Linh Pham, Dung Vo, Thai Tran, Mai Nguyen, Tri Nguyen, Vu Le·· 16 小时前
基于预训练文本到文本模型与 Read-Tag-Write 策略的高效流式逆文本正则化
Towards Efficient Simultaneous Inverse Text Normalization with Pretrained Text-to-Text Language Model and Read-Tag-Write Policy
AI 导读
研究者提出一种高效的流式端到端逆文本正则化（ITN）系统，通过适配预训练文本到文本模型，结合架构调整、专门训练策略、Read-Tag-Write 解码策略与推理优化实现流式处理。在越南语数据集上，其准确率与非流式基线相当，优于混合方法，并满足实时延迟要求。该工作发表于 Interspeech 2026 长文赛道。
正文
Abstract:Streaming Inverse Text Normalization (ITN) is vital for converting spoken-form outputs from streaming Automatic Speech Recognition into formatted written text. Existing streaming ITN methods rely on hybrid systems combining neural tagging with expert-crafted finite-state transducer rules, limiting scalability across domains and languages. While end-to-end models offer superior scalability by learning directly from data, standard encoder-decoder architectures are inherently non-streaming due to global attention. In this paper, we propose an efficient streaming end-to-end ITN system, adapting a pretrained text-to-text model to leverage its robust linguistic knowledge. To enable streaming, we introduce architectural adaptations, a specialized training strategy, a Read-Tag-Write decoding policy, and inference optimizations. Experiments on Vietnamese datasets show accuracy comparable to non-streaming baselines, outperforming hybrid methods while satisfying real-time latency requirements.
| Comments: | Interspeech 2026 [Long Track], 9 pages |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.19492 [cs.CL] |
| (or arXiv:2610.19492v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.19492 arXiv-issued DOI via DataCite (pending registration) |
|
| Related DOI: | https://doi.org/10.21437/Interspeech.2026-1060
DOI(s) linking to related resources |
Submission history
From: Khanh Le Duy [view email]
[v1]
Wed, 14 Oct 2026 03:13:23 UTC (1,038 KB)
来源：arXiv：cs.CL · arxiv.org