English

ConvFill: Model Collaboration for Responsive Conversational Voice Agents

Computation and Language 2025-11-11 v1

Abstract

Deploying conversational voice agents with large language models faces a critical challenge: cloud-based foundation models provide deep reasoning and domain knowledge but introduce latency that disrupts natural conversation, while on-device models respond immediately but lack sophistication. We propose conversational infill, a task where a lightweight on-device model generates contextually appropriate dialogue while seamlessly incorporating streaming knowledge from a powerful backend model. This approach decouples response latency from model capability, enabling systems that feel responsive while accessing the full power of large-scale models. We present ConvFill, a 360M parameter model trained on synthetic multi-domain conversations. Evaluation across multiple backend models shows that conversational infill can be successfully learned, with ConvFill achieving accuracy improvements of 36-42% over standalone small models of the same size while consistently retaining sub-200ms response latencies. Our results demonstrate the promise of this approach for building on-device conversational agents that are both immediately responsive and knowledgeable.

Keywords

Cite

@article{arxiv.2511.07397,
  title  = {ConvFill: Model Collaboration for Responsive Conversational Voice Agents},
  author = {Vidya Srinivas and Zachary Englhardt and Maximus Powers and Shwetak Patel and Vikram Iyer},
  journal= {arXiv preprint arXiv:2511.07397},
  year   = {2025}
}
R2 v1 2026-07-01T07:30:23.020Z