Pardon? Evaluating Conversational Repair in Large Audio-Language Models
This work addresses a reliability gap in conversational AI for real-world interactions by exposing limitations in current accuracy-centric evaluations.
The paper introduced a repair-aware evaluation setting for Large Audio-Language Models (LALMs) to assess their ability to distinguish between answerable and unanswerable spoken inputs, revealing that while models perform well on answerable inputs, most fail to recognize unanswerability and initiate appropriate conversational repair.
Large Audio-Language Models (LALMs) have demonstrated strong performance in spoken question answering (QA), with existing evaluations primarily focusing on answer accuracy and robustness to acoustic perturbations. However, such evaluations implicitly assume that spoken inputs remain semantically answerable, an assumption that often fails in real-world interaction when essential information is missing. In this work, we introduce a repair-aware evaluation setting that explicitly distinguishes between answerable and unanswerable audio inputs. We define answerability as a property of the input itself and construct paired evaluation conditions using a semantic-acoustic masking protocol. Based on this setting, we propose the Evaluability Awareness and Repair (EAR) score, a non-compensatory metric that jointly evaluates task competence under answerable conditions and repair behavior under unanswerable conditions. Experiments on two spoken QA benchmarks across diverse LALMs reveal a consistent gap between answer accuracy and conversational reliability: while many models perform well when inputs are answerable, most fail to recognize semantic unanswerability and initiate appropriate conversational repair. These findings expose a limitation of prevailing accuracy-centric evaluation practices and motivate reliability assessments that treat unanswerable inputs as cues for repair and continued interaction.