SPUR: A Plug-and-Play Framework for Integrating Spatial Audio Understanding and Reasoning into Large Audio-Language Models

TL;DR

SPUR is a lightweight framework that enhances large audio-language models with spatial perception capabilities, enabling better understanding of acoustic scenes by integrating spatial cues through minimal architectural modifications.

Contribution

We introduce SPUR, a plug-in method that equips LALMs with spatial awareness using a FOA encoder and a new spatial QA dataset, improving spatial reasoning without extensive retraining.

Findings

Enhanced spatial QA performance on SPUR-Set

Improved multi-speaker attribution accuracy

Preserved general audio understanding after fine-tuning

Abstract

Spatial perception is central to auditory intelligence, enabling accurate understanding of real-world acoustic scenes and advancing human-level perception of the world around us. While recent large audio-language models (LALMs) show strong reasoning over complex audios, most operate on monaural inputs and lack the ability to capture spatial cues such as direction, elevation, and distance. We introduce SPUR, a lightweight, plug-in approach that equips LALMs with spatial perception through minimal architectural changes. SPUR consists of: (i) a First-Order Ambisonics (FOA) encoder that maps (W, X, Y, Z) channels to rotation-aware, listener-centric spatial features, integrated into target LALMs via a multimodal adapter; and (ii) SPUR-Set, a spatial QA dataset combining open-source FOA recordings with controlled simulations, emphasizing relative direction, elevation, distance, and overlap for supervised spatial reasoning. Fine-tuning our model on the SPUR-Set consistently improves spatial QA and multi-speaker attribution while preserving general audio understanding. SPUR provides a simple recipe that transforms monaural LALMs into spatially aware models. Extensive ablations validate the effectiveness of our approach.