Metraipolis › archive › dc_ff7e64e2c708
SpeechCritic: Learning a Diagnostic Speech Judge from Limited Human Preferences — arXiv:2609.34582
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.34582
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T20:57:57.925Z
- SHA-256 of the archived text
37c5aa1566f31dc45a3297216c5cfbbb49b20eec49a536038bc31b3ca5e0508c- Stable URL of this page
https://metraipolis.com/documento/dc_ff7e64e2c708
How to cite
SpeechCritic: Learning a Diagnostic Speech Judge from Limited Human Preferences — arXiv:2609.34582 — arXiv (en), https://arxiv.org/abs/2609.34582 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ff7e64e2c708 (Metraipolis archive dc_ff7e64e2c708, sha256 37c5aa1566f31dc45a3297216c5cfbbb49b20eec49a536038bc31b3ca5e0508c)
BibTeX
@misc{dcff7e64e2c708,
title = {SpeechCritic: Learning a Diagnostic Speech Judge from Limited Human Preferences — arXiv:2609.34582},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.34582},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ff7e64e2c708; sha256 37c5aa1566f31dc45a3297216c5cfbbb49b20eec49a536038bc31b3ca5e0508c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34582. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Human speech conveys rich perceptual information, such as emotion and speaker identity, yet most automatic speech quality judges reduce it to a single naturalness score. We study diagnostic speech judges: given two candidates, a diagnostic judge decides which is better, along which perceptual dimensions (e.g., timbre, emotion, timing) they differ, and which audible cues support its decision. Learning such judges is challenging: expert annotation is costly, and simply prompting a frontier audio-language model to produce labels is unreliable: our probing reveals substantial errors and unstable instruction following. We introduce SpeechCritic, which learns a diagnostic judge in a reference-conditioned cross-lingual setting from only about 300 human-labeled comparisons. Rather than replacing the frontier model, SpeechCritic calibrates it with these labels: for each dimension, it selects the acoustic measurements that agree with human judgments, maps them to A/Tie/B probabilities, and passes these to the model as non-binding hints alongside the audio. Compared with the same model labeling without hints, this raises dimension-level agreement with humans by 6.3 points and cuts the mismatch with human Tie rates by 10.4 points. We then train a 7B judge on this supervision and find that different training signals shape different judge behaviors: SFT establishes the task, OPD transfers the teacher's dimension-level strengths and weaknesses, and RL helps most on clear-cut comparisons where human raters agree. Notably, human listeners also find that RL makes rationales cite more specific, localized acoustic cues, although it never directly rewards rationale text. Finally, we show that the pipeline is language-pair agnostic by instantiating it on both English-Japanese and English-Spanish. Together, these results demonstrate a path from limited human preferences to a diagnostic speech judge. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI) ; Sound (cs.SD). Pagina del lavoro: https://arxiv.org/abs/2609.34582 — 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 2026-10-07T20:29:01.600Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).