SAKURA: On the Multi-hop Reasoning of Large Audio-Language Models Based on Speech and Audio Information

TL;DR

SAKURA is a new benchmark designed to evaluate the multi-hop reasoning abilities of large audio-language models, revealing their current struggles in integrating speech and audio information for complex reasoning tasks.

Contribution

The paper introduces SAKURA, the first benchmark specifically assessing multi-hop reasoning in large audio-language models, highlighting a key limitation in their multimodal reasoning capabilities.

Findings

LALMs struggle with multi-hop reasoning despite correct information extraction.

Current models have a fundamental challenge in integrating speech/audio for reasoning.

SAKURA exposes critical limitations, guiding future research directions.

Abstract

Large audio-language models (LALMs) extend the large language models with multimodal understanding in speech, audio, etc. While their performances on speech and audio-processing tasks are extensively studied, their reasoning abilities remain underexplored. Particularly, their multi-hop reasoning, the ability to recall and integrate multiple facts, lacks systematic evaluation. Existing benchmarks focus on general speech and audio-processing tasks, conversational abilities, and fairness but overlook this aspect. To bridge this gap, we introduce SAKURA, a benchmark assessing LALMs' multi-hop reasoning based on speech and audio information. Results show that LALMs struggle to integrate speech/audio representations for multi-hop reasoning, even when they extract the relevant information correctly, highlighting a fundamental challenge in multimodal reasoning. Our findings expose a critical limitation in LALMs, offering insights and resources for future research.