Metraipolis › archive › dc_368e0a99ffdb
T-SANDHI: Tone Sandhi-aware Adaptive Network with Decoupled Hybrid Injection for Low-resource Taiwanese Hokkien Speech Recognition — arXiv:2609.18194
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.18194
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T13:49:21.805Z
- SHA-256 of the archived text
24de4e4414bc07b9f85e3e6c8df8dfd83c3fa9ffe44b3c184e4a790bff982499- Stable URL of this page
https://metraipolis.com/documento/dc_368e0a99ffdb
How to cite
T-SANDHI: Tone Sandhi-aware Adaptive Network with Decoupled Hybrid Injection for Low-resource Taiwanese Hokkien Speech Recognition — arXiv:2609.18194 — arXiv (en), https://arxiv.org/abs/2609.18194 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_368e0a99ffdb (Metraipolis archive dc_368e0a99ffdb, sha256 24de4e4414bc07b9f85e3e6c8df8dfd83c3fa9ffe44b3c184e4a790bff982499)
BibTeX
@misc{dc368e0a99ffdb,
title = {T-SANDHI: Tone Sandhi-aware Adaptive Network with Decoupled Hybrid Injection for Low-resource Taiwanese Hokkien Speech Recognition — arXiv:2609.18194},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.18194},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_368e0a99ffdb; sha256 24de4e4414bc07b9f85e3e6c8df8dfd83c3fa9ffe44b3c184e4a790bff982499},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.18194. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
In Taiwanese Hokkien automatic speech recognition (ASR), prior studies often treat tone sandhi as a major challenge under the assumption that models fail to process implicit phonological variations. However, our experiments on Taiwanese Hokkien reveal that speech foundation models actually handle tone sandhi variations effectively, and the real performance bottleneck stems from a localized confusion between these variations and retained citation tones. To address this, we propose T-SANDHI to explicitly decouple surface acoustics from underlying lexical intent on top of a frozen Whisper backbone. Using a lexicon-guided multi-task learning structure driven by text-derived pseudo labels, our lightweight hybrid injection module integrates independent citation and sandhi phonetic streams via dynamic gating. Extensive evaluation on the TAT-MOE corpus and two blind test sets demonstrates that this explicit disentanglement effectively resolves tonal mapping confusion, outperforming baselines with strict parameter efficiency. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-17 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.18194 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-15T13:48:24.302Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-17?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).