Metraipolis › archive › dc_db3378fad5ce

Diagnosing and Mitigating Long-Session Degradation in Full-Duplex Speech Models — arXiv:2610.34991

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.34991
License (as archived)
CC0-1.0
Archived
2026-10-27T08:11:42.193Z
SHA-256 of the archived text
29afe1836fd366fae9d8b68af8be1eb6d3a7624cabdee835522dd0cad536809e
Stable URL of this page
https://metraipolis.com/documento/dc_db3378fad5ce

How to cite

Diagnosing and Mitigating Long-Session Degradation in Full-Duplex Speech Models — arXiv:2610.34991 — arXiv (en), https://arxiv.org/abs/2610.34991 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_db3378fad5ce (Metraipolis archive dc_db3378fad5ce, sha256 29afe1836fd366fae9d8b68af8be1eb6d3a7624cabdee835522dd0cad536809e)
BibTeX
@misc{dcdb3378fad5ce,
  title = {Diagnosing and Mitigating Long-Session Degradation in Full-Duplex Speech Models — arXiv:2610.34991},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.34991},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_db3378fad5ce; sha256 29afe1836fd366fae9d8b68af8be1eb6d3a7624cabdee835522dd0cad536809e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.34991. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

End-to-end full-duplex spoken dialogue models enable natural, real-time spoken interactions with interruptions, backchannels, and overlapping speech. Prior evaluations have assessed diverse interaction capabilities and identified some degradation patterns across conversational turns. However, how model behavior evolves over extended sessions remains insufficiently understood. In this work, we address this gap by systematically characterizing long-session degradation and developing efficient methods to mitigate it. We introduce a closed-loop evaluation framework for long-session stability and apply it to several publicly available full-duplex speech models, revealing degradation over time in responsiveness, spoken content quality, and voice consistency. To mitigate this degradation, we propose a self-distillation-based training framework for lightweight LoRA adaptation that improves long-session stability with a small amount of additional training. We further introduce a two-resolution context (TRC) architecture that preserves recent interactions in full detail while compressing older interaction history into key-value representations. We evaluate our methods on three models with distinct architectures, assessing long-session stability, information retention, and computational efficiency. Across all three models, we show that LoRA adaptation mitigates performance degradation over ten-minute sessions, while adding TRC extends these gains beyond the training session length and reduces KV-cache growth by more than 90%.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.34991 — 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 2026-10-27T07:56:00.245Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).