ConvFill: Model Collaboration for Responsive Conversational Voice Agents

• URL: http://arxiv.org/abs/2511.07397v1
• Date: Mon, 10 Nov 2025 18:50:30 GMT
• Title: ConvFill: Model Collaboration for Responsive Conversational Voice Agents
• Authors: Vidya Srinivas, Zachary Englhardt, Maximus Powers, Shwetak Patel, Vikram Iyer,
• Abstract summary: 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.<n>We present ConvFill, a 360M parameter model trained on synthetic multi-domain conversations.<n>We show 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.
• Score: 6.166061057506208
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• 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.

Related papers

• AMUSE: Audio-Visual Benchmark and Alignment Framework for Agentic Multi-Speaker Understanding [73.05946667683259]
  Recent large language models (MLLMs) show strong perception but struggle in multi-speaker, dialogue-centric settings.<n>We introduce AMUSE, a benchmark designed around tasks that are inherently agentic.<n>We propose RAFT, a data-efficient agentic alignment framework that integrates reward optimization with intrinsic multimodal self-evaluation.
  arXiv  Detail & Related papers  (2025-12-18T07:01:47Z)
• Audio MultiChallenge: A Multi-Turn Evaluation of Spoken Dialogue Systems on Natural Human Interaction [12.216811577733125]
  We introduce Audio MultiChallenge, an open-source benchmark to evaluate E2E spoken dialogue systems under natural multi-turn interaction patterns.<n>We introduce a new axis Voice Editing that tests robustness to mid-utterance speech repairs and backtracking.<n>We curate 452 conversations from 47 speakers with 1,712 instance-specific rubrics through a hybrid audio-native agentic and human-in-the-loop pipeline.
  arXiv  Detail & Related papers  (2025-12-16T19:26:44Z)
• DiscussLLM: Teaching Large Language Models When to Speak [9.441455921296301]
  Large Language Models (LLMs) have demonstrated remarkable capabilities in understanding and generating human-like text.<n>We introduce $textitDiscussLLM$, a framework designed to bridge this gap by training models to proactively decide not just $textitwhat$ to say, but critically, $textitwhen$ to speak.
  arXiv  Detail & Related papers  (2025-08-25T16:16:42Z)
• Vision-Speech Models: Teaching Speech Models to Converse about Images [67.62394024470528]
  We introduce MoshiVis, augmenting a recent dialogue speech LLM, Moshi, with visual inputs through lightweight adaptation modules.<n>An additional dynamic gating mechanism enables the model to more easily switch between the visual inputs and unrelated conversation topics.<n>We evaluate the model on downstream visual understanding tasks with both audio and text prompts, and report qualitative samples of interactions with MoshiVis.
  arXiv  Detail & Related papers  (2025-03-19T18:40:45Z)
• Full-Duplex-Bench: A Benchmark to Evaluate Full-duplex Spoken Dialogue Models on Turn-taking Capabilities [93.09944267871163]
  FullDuplexBench is a benchmark that systematically evaluates key interactive behaviors.<n>By releasing our benchmark code we aim to advance spoken dialogue modeling and the development of more natural and engaging SDMs.
  arXiv  Detail & Related papers  (2025-03-06T18:59:16Z)
• Mini-Omni: Language Models Can Hear, Talk While Thinking in Streaming [0.0]
  Mini- Omni is an audio-based end-to-end conversational model capable of real-time speech interaction. We propose a text-instructed speech generation method, along with batch-parallel strategies during inference to boost the performance. We also introduce the VoiceAssistant-400K dataset to fine-tune models for optimized speech output.
  arXiv  Detail & Related papers  (2024-08-29T17:18:53Z)
• Pheme: Efficient and Conversational Speech Generation [52.34331755341856]
  We introduce the Pheme model series that offers compact yet high-performing conversational TTS models. It can be trained efficiently on smaller-scale conversational data, cutting data demands by more than 10x but still matching the quality of the autoregressive TTS models.
  arXiv  Detail & Related papers  (2024-01-05T14:47:20Z)
• Improving a sequence-to-sequence nlp model using a reinforcement learning policy algorithm [0.0]
  Current neural network models of dialogue generation show great promise for generating answers for chatty agents. But they are short-sighted in that they predict utterances one at a time while disregarding their impact on future outcomes. This work commemorates a preliminary step toward developing a neural conversational model based on the long-term success of dialogues.
  arXiv  Detail & Related papers  (2022-12-28T22:46:57Z)
• GODEL: Large-Scale Pre-Training for Goal-Directed Dialog [119.1397031992088]
  We introduce GODEL, a large pre-trained language model for dialog. We show that GODEL outperforms state-of-the-art pre-trained dialog models in few-shot fine-tuning setups. A novel feature of our evaluation methodology is the introduction of a notion of utility that assesses the usefulness of responses.
  arXiv  Detail & Related papers  (2022-06-22T18:19:32Z)
• Plug-and-Play Conversational Models [62.77150879036442]
  We introduce an approach that does not require further computation at decoding time, while also does not require any fine-tuning of a large language model. We demonstrate, through extensive automatic and human evaluation, a high degree of control over the generated conversational responses with regard to multiple desired attributes.
  arXiv  Detail & Related papers  (2020-10-09T03:17:51Z)
• The Adapter-Bot: All-In-One Controllable Conversational Model [66.48164003532484]
  We propose a dialogue model that uses a fixed backbone model such as DialGPT and triggers on-demand dialogue skills via different adapters. Depending on the skills, the model is able to process multiple knowledge types, such as text, tables, and emphatic responses. We evaluate our model using automatic evaluation by comparing it with existing state-of-the-art conversational models.
  arXiv  Detail & Related papers  (2020-08-28T10:59:31Z)
• Speaker Sensitive Response Evaluation Model [17.381658875470638]
  We propose an automatic evaluation model based on the similarity of the generated response with the conversational context. We learn the model parameters from an unlabeled conversation corpus. We show that our model can be applied to movie dialogues without any additional training.
  arXiv  Detail & Related papers  (2020-06-12T08:59:10Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.