Metraipolis › archive › dc_678148edbe91

Group Relative Policy Optimisation Improves Multilingual Speech Recognition in Low-Resource Languages — arXiv:2610.30663

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.30663
License (as archived)
CC0-1.0
Archived
2026-10-23T15:12:57.873Z
SHA-256 of the archived text
5a34d562b8965518d8135affea48e5127dd9a4e5a0dd34a4a039393581a016dd
Stable URL of this page
https://metraipolis.com/documento/dc_678148edbe91

How to cite

Group Relative Policy Optimisation Improves Multilingual Speech Recognition in Low-Resource Languages — arXiv:2610.30663 — arXiv (en), https://arxiv.org/abs/2610.30663 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_678148edbe91 (Metraipolis archive dc_678148edbe91, sha256 5a34d562b8965518d8135affea48e5127dd9a4e5a0dd34a4a039393581a016dd)
BibTeX
@misc{dc678148edbe91,
  title = {Group Relative Policy Optimisation Improves Multilingual Speech Recognition in Low-Resource Languages — arXiv:2610.30663},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.30663},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_678148edbe91; sha256 5a34d562b8965518d8135affea48e5127dd9a4e5a0dd34a4a039393581a016dd},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.30663. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Autoregressive encoder-decoder speech recognition models trained with teacher forcing can suffer from exposure bias, limiting their ability to recover from their own prediction errors at inference time. Recent work has shown that reinforcement learning with Group Relative Policy Optimisation (GRPO) improves speech recognition in major languages with LLM-based models. We extend this line of work to smaller, non-LLM encoder-decoder ASR models, using Whisper-large-v3 as a case study across 51 African languages. The benefits of this approach are most pronounced when training data is scarce or when input audio is noisy or degraded. Our resulting model, SunflowerASR, achieves the best published macro-averaged WER on the AfriVox-v2 (0.227) and SimbaBench (0.372) benchmarks, and is publicly available.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.30663 — 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 2026-10-23T15:09:08.383Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).