Metraipolis › archive › dc_256ecf7ce950
Do Audio LLMs Listen Before They Act? Diagnosing Acoustic-Context Gating in Voice Agents — arXiv:2609.32536
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.32536
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T21:00:22.296Z
- SHA-256 of the archived text
209b0ced3ab482d82bb5447a81635dc716a0b3a8f4c95bd7f959962387a07031- Stable URL of this page
https://metraipolis.com/documento/dc_256ecf7ce950
How to cite
Do Audio LLMs Listen Before They Act? Diagnosing Acoustic-Context Gating in Voice Agents — arXiv:2609.32536 — arXiv (en), https://arxiv.org/abs/2609.32536 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_256ecf7ce950 (Metraipolis archive dc_256ecf7ce950, sha256 209b0ced3ab482d82bb5447a81635dc716a0b3a8f4c95bd7f959962387a07031)
BibTeX
@misc{dc256ecf7ce950,
title = {Do Audio LLMs Listen Before They Act? Diagnosing Acoustic-Context Gating in Voice Agents — arXiv:2609.32536},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.32536},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_256ecf7ce950; sha256 209b0ced3ab482d82bb5447a81635dc716a0b3a8f4c95bd7f959962387a07031},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32536. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Audio language models can recognize spoken commands and invoke tools, but an agent must first decide whether the acoustic and conversational context warrants action. We introduce VGBench, a 1,018-item diagnostic benchmark for action-level addressedness across side-talk, self-talk, and speaker-switch scenarios. Each item uses a shared action space comprising silence, a tool call, and a natural-language answer. Speaker-switch pairs hold the specified words fixed while source, distance rendering, and a temporal boundary define a controlled wearer-to-bystander shift. Six raw Audio LLMs and three training-free adaptations often identify the target tool yet rarely withhold action under this shift; the highest raw switch mute rate is 14%. We then use VoxGate as a post-training case study. Supervised training mutes 91.3% of switched commands while choosing the correct tool for all nearby wearer commands and text-only controls. An exploratory GRPO stage has similar switch performance; side-talk accuracy rises from 68.4% to 70.9%, and self-talk muting from 52.0% to 60.0%. Factorized controls identify an independent source-change effect, while sensitivity to the far-field manipulation varies across acoustic renderings. The benchmark therefore measures multi-cue acoustic-context gating rather than isolated speaker identity. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI). Categorie arXiv: Sound (cs.SD) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multimedia (cs.MM); Audio and Speech Processing (eess.AS). Pagina del lavoro: https://arxiv.org/abs/2609.32536 — 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 2026-10-07T20:29:02.378Z 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).