CS Paper Radar

Hiding Tool Latency in On-Device Cascaded Voice Agent through Speculative Execution

Kyudan Jung, Hyunsin Park, Yoonhyung Lee, Jinhwan Park, Jinhyeok Yang, KiHyun Nam, Jaegul Choo, Jinkyu Lee

cs.SD ·submitted 2026-10-06 02:36 UTC · v1

Abstract

Tool-augmented speech assistants typically serialize automatic speech recognition, large language model inference, and external tool execution. As a result, tool latency is incurred only after the user has finished speaking and the LLM has identified the required tool calls. We present speculative tool execution for on-device cascaded voice agents, which predicts tool requests from partial ASR hypotheses and initiates tool execution while speech is still being received, thereby reducing end-to-end response latency. Our approach introduces a Predictor module that anticipates tool calls during speech recognition, executes them speculatively, and caches the results. The cached outputs are then injected into the LLM prompt, enabling faster responses. Additionally, to mitigate errors caused by user self-corrections during speech, we employ a rule-based validation mechanism that selectively injects only valid cached results. As a final safeguard, the LLM retains the ability to issue tool calls directly, ensuring that the latency of our framework is upper-bounded by the baseline serial execution pipeline in the worst case. We evaluate our method using live measurements from a fully implemented Android voice assistant. Our approach reduces the median time-to-first-audio from 5.79,s to 4.60,s and decreases the standard deviation from 3.49,s to 2.81,s, resulting in more predictable response latency.

Comment
5 pages, 2 figures, 4 tables

Closest papers

Nearest neighbours by meaning across the whole index.

SURE-EVAL: A Systematic and Unified Agentic Framework for Reproducible Evaluation

eess.AS ·2026-09-30 ·Jing Peng, Junhao Du, Yixuan Wang, Bowen Wang, Hanqi Li +19 more
publishedcode/data similarity 0.71
Abstract · This paper is accecpted to NCMMSC2026 as an oral presentation

Audio and speech models are released rapidly, but reported scores often conflate model capability with deployment and evaluation choices. The same checkpoint can produce different predictions under different runtimes, hardware, decoding settings, or fallback policies. Even fixed predictions can receive different scores under different normalization and metric implementations. Existing speech benchmarks standardize selected datasets or scoring procedures, but rarely connect heterogeneous model onboarding, controlled inference, and versioned scoring in one executable workflow. We introduce SURE-EVAL, a Systematic and Unified Agentic framework for Reproducible Evaluation of audio and speech systems. A Tool Agent Workflow converts model releases into isolated, verified callable tools. A Main Agent Workflow commits task-specific inference and scoring protocols, then delegates all score-bearing operations to versioned deterministic programs. Each result retains its runtime, protocol, pipeline nodes, predictions, and audit artifacts. Across 18 public releases covering automatic speech recognition, text-to-speech, voice conversion, speaker diarization, speaker-attributed recognition, and multi-task audio understanding, a Codex-only baseline completes 12 models in one shot, while the same agent with the SURE-EVAL Tool Agent Workflow completes all 18. We also conduct unified evaluations over seven ASR test conditions and two TTS subsets. A protocol analysis of three TTS systems finds absolute differences of 0.02-0.52 points between paper-reported and unified results, with the direction varying by model and language. These results show that reproducible evaluation requires controlling both model execution and output scoring.

0.30

Talk2Agent: Benchmarking Voice Interfaces for Text Agents

cs.AI eess.AS ·2026-09-30 ·Terumi Chiba, Guangzhi Sun, Zheqi Yuan, Chao Zhang
benchmark similarity 0.71
Abstract

Large language model (LLM) computer-use agents are typically evaluated with clean written instructions, despite speech being an increasingly popular interface for interacting with such systems. Speech input introduces an additional failure point: transcription errors can alter task-critical entities, constraints, or targets before the agent begins reasoning, while conventional ASR metrics do not directly measure whether the information required for successful execution has been preserved. We introduce Talk2Agent, a benchmark for evaluating how effectively voice interfaces convey human-spoken instructions to LLM-based computer-use agents. Talk2Agent builds human-spoken versions of tasks from WildClawBench and OSWorld and evaluates a range of voice interfaces, including dedicated ASR models, audio-capable LLMs, contextual biasing, and LLM-based ontology repair. Because repeatedly executing long-horizon computer-use tasks is costly and stochastic, we further propose an execution-free, task-conditioned evaluation framework that projects the original task grader onto prompt-addressable intentions and measures how much task-relevant information is retained after the voice interface. On WildClawBench, Talk2Agent's execution-free native projection provides a practical, execution-grounded measure of voice-interface quality, correlating with downstream task completion and improving Pearson correlation by 0.246 over WER/CER on 32 hours of real human speech.

0.12

Heartbeat Thinking: Speculative Reasoning and Tool Execution While Listening in Realtime Voice Agents

cs.CL ·2026-10-19 ·Zhenyu Zhang, Kewei Fu, Yongqi Tong, Chenzhi Zheng, Wei Wu +2 more
benchmark similarity 0.71
Abstract

Tool-using voice agents often leave users waiting after speech ends, even when actionable information arrives earlier in the request. Speculative approaches begin work during speech, but chunk-bounded or interrupt-on-update generation can truncate a tool call before its arguments are complete. This couples computation time to speech updates, limiting useful work completed in advance. We introduce Heartbeat Thinking, which handles updated input through new Thinker invocations while allowing ongoing invocations and dispatched tools to finish. Periodic invocations read the latest partial transcript and a shared record of completed reasoning and tool results. These results help resolve missing arguments and initiate dependent calls, allowing preparation to progress across speech updates. Across SpeechFCEval, FDBench v3, and AudioCRAG, Heartbeat improves aggregate speculative tool coverage or retrieval reuse over a SHANKS-style windowed baseline, with benchmark-dependent effects on final task quality. On FDBench v3, it increases tool coverage by 16 percentage points and reduces mean final-response latency by 36.4\%. The results show that heartbeat thinking paradigm preserving ongoing computation during speech can turn listening time into reusable preparation for faster and more capable voice-agent responses.

0.22

A frontend-backend architecture for tool calls in full-duplex speech models

cs.CL ·2026-09-16 ·Ke Hu, Slyne Deng, Chen Chen, Elena Rastorgueva, Edresson Casanova +8 more
highly cited similarity 0.70
Abstract · To be submitted to ICASSP'27

Full-duplex speech-to-speech (S2S) models provide natural, low-latency conversational interaction and would benefit from the ability to use external tools and complete voice-agent tasks. We propose a frontend-backend architecture where a duplex speech-to-text frontend learns to emit a delegation token and forwards streaming ASR transcripts to a text-based backend LLM for tool calls. Tool-call results from the backend are injected back into the frontend through a lightweight prefill-and-repeat mechanism and then synthesized using streaming TTS to the user. Our approach largely preserves regular duplex turn-taking, interruption handling, and low-latency interaction as it requires minimal modifications to the frontend model. In a single-turn tool-call evaluation, our system achieves 92-97% tool-call recall, competitive tool-call prediction performance, and 81.2% accuracy in rejecting irrelevant calls. When equipped with a larger backend (e.g., Qwen3-235B-A22B), our system achieves competitive results on Full-Duplex-Bench-V3 compared to open and closed source models, and significantly outperforms GPT-realtime-mini and Qwen3-Omni-30B-A3B-Instruct on EVA-Bench. These results demonstrate that backend delegation is an effective and modular approach for combining natural duplex speech interaction with strong agentic tool-call capabilities.

0.44

MTVA-Bench: Evaluating the Language Model Inside Cascaded Voice Agents

cs.AI ·2026-09-17 ·Pritish Mishra, Ishaan Kumar, Akshat Mandoli, Sudarshan Kamath
benchmark similarity 0.69
Abstract

Generally, most voice agents are cascaded systems, i.e., an ASR model transcribes the caller's audio, a language model reads the transcript and decides what to say and which backend tools to call, and a TTS model speaks the reply. Nearly all of the decision making happens in the language model, but existing evaluations measure it either too broadly or too narrowly. End-to-end voice benchmarks score the full pipeline, so recognition errors and model errors mix into a single number. LLM benchmarks isolate the model but they do not evaluate what makes real phone calls hard, such as transcription issues, caller's voice being split across messages and the requirement that replies follow the language and script specified. We introduce the Multi-Turn Voice Agent Benchmark (MTVA-Bench), which evaluates the language model on the same conditions it faces inside a cascaded system. The caller is played by an LLM following a set of rubrics and tool calls are answered by a mock backend which responds to the arguments the model actually sent. The benchmark contains 49 agents working across 490 reviewed scenarios and supports 7 languages. Scoring is a combination of deterministic checks on tool calls with two LLM judges, one that scores scenario specific rules and one that grades conversation quality without access to the task. Both judges must cite specific messages from the transcript. Task and conversation scores are weighted equally, since a call can complete its task and still go badly for the caller. In a seven-model study, six of the models select the correct tool within 6.4 points of one another, but their overall scores span 24.4 points. Most of the gap comes from argument values, action ordering, rule compliance, and what the model says around its tool calls.

0.48

Scalable Context Orchestration for Serving LLMs Over Voice

cs.SD cs.AIeess.AS ·2026-09-03 ·Linyi Jiang, Silvery D. Fu, Yifei Zhu
top venue similarity 0.68
Abstract · Accepted for publication in ACM SOSP 2026

Voice AI applications are gaining popularity as advances in large language models (LLMs) enable more natural and accessible spoken interactions. Serving these applications requires accounting not only for what users say, but also for how they speak (e.g., speaking rate) and the conditions under which their audio is captured and transmitted (e.g., background noise and packet loss). However, existing LLM systems represent conversation context as a flat, growing sequence of messages, leaving voice-specific context implicit in the audio. As a result, they can generate responses that are poorly aligned with user preferences, degrade interaction quality under adverse environmental conditions, and incur high costs over long voice sessions. We present llmovoice, a context-management middleware that explicitly models voice context and orchestrates its use. At each turn, llmovoice constructs a bounded voice context from the current user input, relevant interaction history, and explicit paralinguistic and environmental states. It then uses the serving LLM to reason over this context and generate runtime directives that guide how the system responds. We evaluate llmovoice on real-world voice applications and benchmarks. It reduces speaking-rate alignment error by 52.4%, lowers the false-interruption rate from 46.0% to 0.9% under packet loss, and reduces model usage cost by 79.2%. For long sessions, llmovoice reduces per-turn cost by up to 24.9 times while retaining up to 98.7% of baseline answer quality.

0.42

Joi-Bench-Task: Benchmarking Task Coordination in Full-duplex Voice Agents through Interactive Simulation

cs.AI ·2026-11-09 ·Xiangheng He, Mingjun Zhao, Zhehao Fan, Zhengjia Li, Xiaopeng Sun +3 more
benchmark similarity 0.68
Abstract

Full-duplex voice agents enable natural, hands-free interaction, allowing users to refine requests, redirect ongoing work, and seek clarification while the agent is speaking. However, existing voice-agent benchmarks insufficiently isolate coordination between task execution and conversation. We introduce a benchmark that assesses this coordination by examining whether agents track what is being executed, what has been completed, what has been communicated, and what remains to be reported as the conversation evolves. We propose an interactive user simulator that introduces configurable interventions at designated timing in tool execution and spoken result delivery, supporting controllable and reproducible evaluation conditions. We evaluate five leading commercial voice agents along three dimensions: execution responsiveness, coordination under request changes, and coordination under interruptions. Agent evaluations reveal persistent coordination failures. Under request changes, the highest result completion rate is only 55%, indicating difficulty for current voice agents adapting to new requests while preserving earlier obligations. After user interruption, four of the five agents exhibit average information omission rates of at least 23%. These findings highlight task coordination as a central challenge for current voice agents in complex, evolving conversations.

0.16

MobiInfer: Accelerating On-Device GUI Agents with Workload-Aware Inference

cs.OS ·2026-08-11 ·Siming Chen, Erhu Feng, Dahu Feng, Yubin Xia
benchmark similarity 0.68
Abstract

Mobile GUI agents assist users in executing phone tasks through iterative observe-reason-act loops. However, deploy ing multimodal GUI agents directly on smartphones is challenging due to strict end-to-end latency requirements and significant memory overhead. We observe that GUI-agent work loads exhibit recurring workflows and structured I/O: their prompts contain reusable context, while their outputs follow fixed schemas with phase-dependent predictability. Based on this observation, we present MobiInfer, a workload-aware inference engine for on-device multimodal GUI agents. MobiInfer combines trajectory-aware prefix reuse, phase-indexed vocabulary specialization, and schema-guided speculative decoding to eliminate redundant prefill computation and reduce structured-generation overhead. We implement MobiInfer on top of MNN and evaluate it on AndroidWorld and a real world dataset using three real Android devices. Across the four model-benchmark settings, MobiInfer achieves average speedups of 2.6x in prefill and 2.5x in decoding, reduces end-to-end task latency by 2.4x on average, and cuts total deployment storage by 30%, while limiting task-success degradation to less than one percentage point.

0.04