Metraipolis › archive › dc_ea4f44eb9101

Language-Specialized Multi-Teacher On-Policy Distillation for Multilingual LLM-Based ASR — arXiv:2608.03610

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.03610
License (as archived): CC0-1.0
Archived: 2026-10-31T12:31:08.301Z
SHA-256 of the archived text: e08de2f9d3f9c8582b062979a4d9cb78afafa11c157c9734a12198198c9e623d
Stable URL of this page: https://metraipolis.com/documento/dc_ea4f44eb9101

How to cite

Language-Specialized Multi-Teacher On-Policy Distillation for Multilingual LLM-Based ASR — arXiv:2608.03610 — arXiv (en), https://arxiv.org/abs/2608.03610 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ea4f44eb9101 (Metraipolis archive dc_ea4f44eb9101, sha256 e08de2f9d3f9c8582b062979a4d9cb78afafa11c157c9734a12198198c9e623d)

BibTeX

@misc{dcea4f44eb9101,
  title = {Language-Specialized Multi-Teacher On-Policy Distillation for Multilingual LLM-Based ASR — arXiv:2608.03610},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.03610},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ea4f44eb9101; sha256 e08de2f9d3f9c8582b062979a4d9cb78afafa11c157c9734a12198198c9e623d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.03610. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Modern LLM-based ASR systems have established multilingual capability as a standard feature, leveraging large-scale multilingual corpora and LLMs' cross-lingual knowledge to achieve competitive performance across multilingual benchmarks. However, jointly modeling languages with heterogeneous acoustic, phonological, and lexical characteristics inevitably introduces optimization conflicts, undermining language-wise specialization. To address this challenge, we propose Language-Specialized Multi-Teacher On-Policy Distillation (LS-MOPD), which decouples language-specific knowledge acquisition from multilingual capability integration: language-specialized teachers are independently optimized via reinforcement learning (RL), with their expertise then integrated into a generalist multilingual student through language routing and token-level multi-teacher distillation, thereby reducing direct cross-lingual optimization conflicts. We further explore static and dynamic acoustic-prefix configurations to examine how teacher-student prefix consistency influences the efficacy of on-policy distillation. Experiments on benchmarks covering Mandarin, Mandarin subdialects, Cantonese, and English demonstrate that LS-MOPD substantially outperforms RL baselines and surpasses the empirical performance envelope defined by the best-performing RL teachers on nearly all benchmarks, revealing its potential to generalize beyond all teachers in multilingual ASR.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Sound (cs.SD); Audio and Speech Processing (eess.AS).
Pagina del lavoro: https://arxiv.org/abs/2608.03610 — 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 2026-10-31T12:17:37.772Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).