arXiv Computation and Language By Maike Z\"ufle, Peter Pol\'ak, Sefik Emre Eskimez, Jan Niehues, Peter Bell, Ond\v{r}ej Klejch

Controlling Backchannels in Streamable Full-duplex Models

Read the original on arXiv Computation and Language →

The paper introduces a lightweight backchannel head that predicts when a backchannel should begin in full-duplex spoken dialogue models, using the models’ hidden states. When the predicted probability exceeds a tunable threshold, a backchannel is force‑decoded. Experiments on 7B and 1B models show that the head generalizes across scale, aligns with human timing, and produces backchannels that human raters judge as comparable to real ones.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

arXiv Computation and Language
Sep 14

SteerDuplex: Steerable Duplex Speech Dialogue Models

SteerDuplex is a full‑duplex speech dialogue model that can be steered along attributes such as tone, persona, speaking rate, and voice style in response to user instructions. The authors introduce a taxonomy of text‑ and audio‑based steerability, identify gaps in existing models, and fine‑tune a Moshi‑based model with reinforcement learning to improve timing and response continuity. They also present SteerBench, a benchmark of 390 spoken prompts and 1,067 human‑authored rubrics, showing significant gains in audio‑steering pass rates and interruption handling compared to open baselines.

By Utkarsh Tyagi, Ramaneswaran Selvakumar, Advait Gosai, Sonal Kumar, Nikhil Barhate, Isabell Sagar, Steven Li, Miheer Bavare, Daniel Quigley, Fabiola Tapia Carrillo, Jose M Patron E, Diego Mac\'ias Guti\'errez, Paul Song, Ramani Duraiswami, Dinesh Manocha, Yunzhong He
arXiv AI
Sep 4

DuplexSpeechBench-IFEval: Evaluating Implicit Instruction Following in Full-Duplex Voice Agents

DuplexSpeechBench-IFEval (DSB-IFEval) is a new benchmark that evaluates how full‑duplex voice agents follow implicit instructions during real‑time spoken interaction. It contains 1,038 test cases across eight assistant roles and tests five conditioning protocols, measuring floor management with an Instruction Adherence Score (IAS) and persona consistency with a Persona Adherence Score (PAS). Experiments on six speech systems reveal architecture‑dependent trade‑offs, showing that some models are more sensitive to explicit versus persona‑only instructions and that even when following conflicting directives, they struggle to override them under safety conflict.

By Puneet Mathur, Dinesh Manocha
arXiv AI
Jun 9

Liberating LLM Capabilities in Full-Duplex Speech Models

arXiv:2606. 07547v1 Announce Type: cross Abstract: Speech-based large language models are typically constrained to spoken replies, which limits their user-facing outputs to what can be verbalized and suppresses text-native capabilities such as code generation, structured analysis, and multi-step reasoning in realtime interaction, for tasks that require persistent, structured, and inspectable intermediate outputs.

By Luoyuan Zhang, Bokai Xu, Junbo Cui, Weiyue Sun, Yingjing Xu, Hanyu Liu, Yuan Yao