Metraipolis › archive › dc_fb891353799e

myMediWhisper: Construction of Burmese Medical Speech Corpus and Whisper Fine-Tuning for Clinical Dialogue ASR — arXiv:2608.11036

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.11036
License (as archived)
CC0-1.0
Archived
2026-10-19T06:14:24.832Z
SHA-256 of the archived text
3d9065a5d7a29d6888a15646a37e149d2081feeffe3130154b3d2f587acbe7eb
Stable URL of this page
https://metraipolis.com/documento/dc_fb891353799e

How to cite

myMediWhisper: Construction of Burmese Medical Speech Corpus and Whisper Fine-Tuning for Clinical Dialogue ASR — arXiv:2608.11036 — arXiv (en), https://arxiv.org/abs/2608.11036 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fb891353799e (Metraipolis archive dc_fb891353799e, sha256 3d9065a5d7a29d6888a15646a37e149d2081feeffe3130154b3d2f587acbe7eb)
BibTeX
@misc{dcfb891353799e,
  title = {myMediWhisper: Construction of Burmese Medical Speech Corpus and Whisper Fine-Tuning for Clinical Dialogue ASR — arXiv:2608.11036},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.11036},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fb891353799e; sha256 3d9065a5d7a29d6888a15646a37e149d2081feeffe3130154b3d2f587acbe7eb},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.11036. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Although Whisper models benefit from large-scale multilingual pre-training, their performance on Burmese medical speech remains limited. This work presents a Burmese medical speech recognition framework built on a high-quality 28-hour corpus recorded and validated by native speakers. We fine-tune Whisper models using full fine-tuning (FFT) and parameter-efficient fine-tuning (PEFT) with LoRA. To evaluate robustness, we apply waveform- and spectrogram-level data augmentation under controlled noise and simulated room acoustics. While augmentation reduces performance on clean speech, it significantly improves robustness in noisy and reverberant environments across FFT and PEFT settings. Our best-performing system, fully fine-tuned myMediWhisper-Medium without augmentation, achieves a state-of-the-art Word Error Rate (WER) of 23.44%, outperforming much larger general-domain fine-tuned models. Dataset and other resources can be found at the Huggingface repository: this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-12 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.11036 — 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 2026-10-19T06:08:50.341Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-12?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).