BAT: Learning to Reason about Spatial Sounds with Large Language Models

TL;DR

BAT combines a novel spatial audio encoder with large language models to enable reasoning about spatial sounds, trained on synthesized datasets, achieving superior perception and reasoning performance in complex audio environments.

Contribution

Introduces Spatial-AST, a new spatial audio encoder, and a comprehensive dataset for training LLMs in spatial sound reasoning tasks.

Findings

BAT outperforms existing models in spatial sound perception tasks.

BAT demonstrates strong reasoning capabilities in complex spatial audio scenarios.

Spatial-AST achieves high accuracy in sound event detection and localization.

Abstract

Spatial sound reasoning is a fundamental human skill, enabling us to navigate and interpret our surroundings based on sound. In this paper we present BAT, which combines the spatial sound perception ability of a binaural acoustic scene analysis model with the natural language reasoning capabilities of a large language model (LLM) to replicate this innate ability. To address the lack of existing datasets of in-the-wild spatial sounds, we synthesized a binaural audio dataset using AudioSet and SoundSpaces 2.0. Next, we developed SpatialSoundQA, a spatial sound-based question-answering dataset, offering a range of QA tasks that train BAT in various aspects of spatial sound perception and reasoning. The acoustic front end encoder of BAT is a novel spatial audio encoder named Spatial Audio Spectrogram Transformer, or Spatial-AST, which by itself achieves strong performance across sound event detection, spatial localization, and distance estimation. By integrating Spatial-AST with LLaMA-2 7B model, BAT transcends standard Sound Event Localization and Detection (SELD) tasks, enabling the model to reason about the relationships between the sounds in its environment. Our experiments demonstrate BAT's superior performance on both spatial sound perception and reasoning, showcasing the immense potential of LLMs in navigating and interpreting complex spatial audio environments.