FuguReport

HEAR Who Said What: Unlocking Speaker-Attributed Reasoning via Counterfactual Voice Grounding

Authors Dongwook Lee, Sangkwon Park, Eunwoo Song, Che Hyun Lee, Youngho Cho, Junho Kim, June Young Yi, Heeseung Kim, Sungroh Yoon
Affiliations Seoul National University / Yonsei University / University of Seoul
Categories Evaluation / Speaker Attribution / Benchmarking speaker attribution abilities, Method / Voice Grounding / Counterfactual voice grounding technique, Application / Speech Language Models / Multi-speaker reasoning and identification
License CC BY 4.0

Abstract Overview

This paper studies whether speech language models can correctly determine who said what in multi-speaker audio and reason over speaker identities. It introduces HEAR, a hierarchical benchmark for speaker-attributed reasoning with 2.4K human-verified multiple-choice samples from 887 real-world audio clips, spanning discrimination, attribution, and reasoning tasks. Evaluations of 20 leading speech language models show that many models perform poorly on foundational speaker attribution and often rely on semantic priors instead of acoustic vocal cues. To address this, the authors build CASH, a 60K-scale counterfactual audio dataset with speaker-level hard negatives, and train A2R, a 30B model that first produces a speaker-tagged transcript and then reasons over it.

Novelty

The main novelty is the explicit framing and diagnosis of speaker-attributed reasoning as a hierarchical capability, operationalized through the new HEAR benchmark. The work is also distinctive in using counterfactual voice grounding via CASH, where the transcript is preserved but the speaker identity is swapped, forcing models to depend on vocal evidence rather than language priors.

Results

On HEAR, A2R outperforms the open-source baselines by a large margin, reaching a 67.1 average score versus 35.4 for its Qwen3-Omni-30B-A3B-Instruct base model, and it attains much stronger paired reasoning accuracy that tests resistance to semantic hallucination. The model also shows zero-shot transfer gains on speaker-attribution-critical tasks, improving over the base model by +36.4 on WDYL, +22.6 on GAOKAO, and +19.0 on FTS. These gains persist on corresponding human-recorded versions of the tasks, while causing only modest degradation on general dyadic VoiceBench evaluations.

Key Points

  1. HEAR provides a structured benchmark for evaluating multi-speaker understanding across discrimination, attribution, and reasoning, with human-verified examples from real-world audio.
  2. The study shows that many current speech language models overestimate their reasoning ability because they often answer from semantic plausibility rather than true voice-grounded speaker attribution.
  3. A2R, trained on the CASH counterfactual dataset with a transcription-first objective, improves speaker-aware reasoning and transfers to unseen attribution-critical tasks, including human-recorded audio.

References

This page was created using generative AI such as GPT-5, Claude Opus 4, Gemini 3, Gemini 3.1 Flash Image, and their higher-end successor versions. No guarantee can be made regarding its contents.