Metraipolis › archive › dc_25bec8d01810
Local or Escalate? Post-Training Expert Routing for Real-Time Duplex Speech Models — arXiv:2611.01639
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.01639
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T18:13:42.349Z
- SHA-256 of the archived text
8b815c7bea799716efe823453e3e229860d6dd033aa82d7a0fd835cf6015515b- Stable URL of this page
https://metraipolis.com/documento/dc_25bec8d01810
How to cite
Local or Escalate? Post-Training Expert Routing for Real-Time Duplex Speech Models — arXiv:2611.01639 — arXiv (en), https://arxiv.org/abs/2611.01639 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_25bec8d01810 (Metraipolis archive dc_25bec8d01810, sha256 8b815c7bea799716efe823453e3e229860d6dd033aa82d7a0fd835cf6015515b)
BibTeX
@misc{dc25bec8d01810,
title = {Local or Escalate? Post-Training Expert Routing for Real-Time Duplex Speech Models — arXiv:2611.01639},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.01639},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_25bec8d01810; sha256 8b815c7bea799716efe823453e3e229860d6dd033aa82d7a0fd835cf6015515b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.01639. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
A full-duplex model should answer easy questions immediately, but hard questions often need stronger reasoning, which may require waiting. The challenge is to choose between these paths at the moment the model begins to speak. Recent systems learn related mode decisions through special-token generation inside the dialogue model, coupling routing to turn-taking and temporally annotated full-duplex training. We study a post-training alternative. We freeze MiniCPM-o 4.5, using only its speech input and output, and fit a logistic failure readout on hidden states already available at its native listen-to-speak transition. The resulting gate requires no additional model generation and exposes thresholds that can be retuned without changing the duplex policy. A probe fitted on text-input states also reads the states of spoken questions, including unseen human recordings. On the internal benchmark, selective handoff improves answer accuracy by 27.1 points. Compared with always using the expert, it reduces mean server time to first answer audio by 31.5 percent. It also outperforms matched random routing across external speech question-answering pools, as well as a zero-shot GPT-5.4 mini router that rates the transcribed question. The training recipe also transfers to a second duplex model family. These results show that an existing duplex model can be equipped to answer easy turns immediately and wait for stronger reasoning or fresher knowledge on the turns that need it, without further duplex training. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI) ; Audio and Speech Processing (eess.AS). Pagina del lavoro: https://arxiv.org/abs/2611.01639 — 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 2026-11-03T18:09:53.347Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).