Beyond Single-Audio: Advancing Multi-Audio Processing in Audio Large Language Models

TL;DR

This paper introduces a multi-audio evaluation benchmark and a novel multi-audio large language model, demonstrating improved multi-audio processing capabilities and data efficiency, advancing towards human-like auditory understanding in machines.

Contribution

It presents the first multi-audio evaluation benchmark and a new multi-audio LLM that effectively captures audio context among multiple streams using synthetic data.

Findings

Existing ALLMs struggle with multi-audio scenarios.

MALLM outperforms baselines in multi-audio tasks.

Synthetic data enables high data efficiency.

Abstract

Various audio-LLMs (ALLMs) have been explored recently for tackling different audio tasks simultaneously using a single, unified model. While existing evaluations of ALLMs primarily focus on single-audio tasks, real-world applications often involve processing multiple audio streams simultaneously. To bridge this gap, we propose the first multi-audio evaluation (MAE) benchmark that consists of 20 datasets from 11 multi-audio tasks encompassing both speech and sound scenarios. Comprehensive experiments on MAE demonstrate that the existing ALLMs, while being powerful in comprehending primary audio elements in individual audio inputs, struggling to handle multi-audio scenarios. To this end, we propose a novel multi-audio-LLM (MALLM) to capture audio context among multiple similar audios using discriminative learning on our proposed synthetic data. The results demonstrate that the proposed MALLM outperforms all baselines and achieves high data efficiency using synthetic data without requiring human annotations. The proposed MALLM opens the door for ALLMs towards multi-audio processing era and brings us closer to replicating human auditory capabilities in machines.