Metraipolis › archive › dc_f43af43758ea

Temporal Taxation Compounds Under Post-Training Compression of Whisper Models — arXiv:2609.28739

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.28739
License (as archived)
CC0-1.0
Archived
2026-10-08T04:49:23.704Z
SHA-256 of the archived text
f12f55e51cc2a4267466ab2b47e42362df6130c4a59cd67660aabd8d40065ec9
Stable URL of this page
https://metraipolis.com/documento/dc_f43af43758ea

How to cite

Temporal Taxation Compounds Under Post-Training Compression of Whisper Models — arXiv:2609.28739 — arXiv (en), https://arxiv.org/abs/2609.28739 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f43af43758ea (Metraipolis archive dc_f43af43758ea, sha256 f12f55e51cc2a4267466ab2b47e42362df6130c4a59cd67660aabd8d40065ec9)
BibTeX
@misc{dcf43af43758ea,
  title = {Temporal Taxation Compounds Under Post-Training Compression of Whisper Models — arXiv:2609.28739},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.28739},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f43af43758ea; sha256 f12f55e51cc2a4267466ab2b47e42362df6130c4a59cd67660aabd8d40065ec9},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.28739. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Automatic speech recognition models are audited for demographic fairness at full precision, yet the models that ship to production have been quantized, pruned, and distilled. We ask whether post-training weight compression, which alters model weights rather than the audio signal or its feature representation, redistributes error burden across demographic groups. Across the Whisper family on Fair-Speech, Common Voice 25, and AfriSpeech-200, 50% Wanda pruning of Whisper-large-v3 sharply widens the Black/AA-vs-Asian temporal-taxation differential on Fair-Speech: the absolute word-error-rate gap between the worst- and best-served groups more than doubles; at an assumed cost of five seconds of correction effort per transcription error this is a rise from 30 to 64 seconds of correction time per minute of speech. This +111% relative increase is invariant to the assumed per-error cost, survives an audio-quality control, and is only partly mitigated by beam-search decoding, which still leaves an +86% increase. At edge model size, INT4 HQQ quantization compounds catastrophic transcript loops on West African accents by factors of five to seven. Distillation, by contrast, narrows demographic gaps in 21 of 27 evaluated settings (teacher-student pair, precision, and dataset), with the exceptions concentrated on a single model pair. We cast the temporal-taxation construct of Choi and Choi (2025) as a quantitative metric, and show that single-snapshot fairness audits on full-precision models do not capture the deployment-time burden that compression places on already-marginalized speakers.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-25 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Sound (cs.SD).
Pagina del lavoro: https://arxiv.org/abs/2609.28739 — 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 2026-10-08T04:27:20.021Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).