yongr/

SpatialAudio

This repo hosts the dataset and models of "BAT: Learning to Reason about Spatial Sounds with Large Language Models" [ICML 2024 bib].

	Spatial Audio Dataset (Mono/Binaural/Ambisonics)




	
	AudioSet (Anechoic Audio Source)

We provide Balanced train and Evaluation set for your convenience. You can download from SpatialAudio. For the Unbalanced train set, please refer to Official AudioSet. Metadata can be downloaded from metadata. AudioSet ├──… See the full description on the dataset page: https://huggingface.co/datasets/yongr/SpatialAudio.