Speech-Hands: A Self-Reflection Voice Agentic Approach to Speech Recognition and Audio Reasoning with Omni Perception

TL;DR

Speech-Hands introduces a self-reflective voice agentic framework that improves speech recognition and audio reasoning by learning when to trust internal models versus external perception, enhancing robustness and accuracy.

Contribution

The paper presents a novel self-reflection mechanism for audio understanding models, significantly improving performance and robustness across speech recognition and audio reasoning tasks.

Findings

Outperforms baselines by 12.1% WER on seven benchmarks

Achieves 77.37% accuracy on audio QA tasks

Generalizes from speech recognition to complex audio reasoning

Abstract

We introduce a voice-agentic framework that learns one critical omni-understanding skill: knowing when to trust itself versus when to consult external audio perception. Our work is motivated by a crucial yet counterintuitive finding: naively fine-tuning an omni-model on both speech recognition and external sound understanding tasks often degrades performance, as the model can be easily misled by noisy hypotheses. To address this, our framework, Speech-Hands, recasts the problem as an explicit self-reflection decision. This learnable reflection primitive proves effective in preventing the model from being derailed by flawed external candidates. We show that this agentic action mechanism generalizes naturally from speech recognition to complex, multiple-choice audio reasoning. Across the OpenASR leaderboard, Speech-Hands consistently outperforms strong baselines by 12.1% WER on seven benchmarks. The model also achieves 77.37% accuracy and high F1 on audio QA decisions, showing robust generalization and reliability across diverse audio question answering datasets. By unifying perception and decision-making, our work offers a practical path toward more reliable and resilient audio intelligence.