FSD50K-Solo: Automated Curation of Single-Source Sound Events

TL;DR

This paper presents FSD50K-Solo, a method for automatically curating a large-scale, high-quality, single-source sound event dataset from open audio corpora using generative models and classifiers.

Contribution

It introduces a novel framework combining generative diffusion models and discriminative classifiers to identify and filter single-source audio samples from large datasets.

Findings

Framework achieves strong performance on a human-curated test set.

FSD50K-Solo contains high-quality single-source audio samples.

Method establishes a scalable paradigm for open-source audio data curation.

Abstract

High-quality training datasets are essential for the performance of neural networks. However, the audio domain still lacks a large-scale, strongly-labeled, and single-source sound event dataset. The FSD50K dataset, despite being relatively large and open, contains a considerable fraction of multi-source samples where background interference or overlapping events could limit the usefulness of the data. To address this challenge, we introduce a data curation framework designed for large-scale open audio corpora. Our approach leverages a generative diffusion model to synthesize clean single-class events to construct controlled noisy mixtures for supervision. We subsequently employ a pre-trained audio encoder coupled with a discriminative classifier to automatically identify and filter out multi-source samples. Experiments show that our framework achieves strong performance on a human expert-curated test set. Finally, we release FSD50K-Solo, a model-curated subset of FSD50K containing single-source audio samples identified by our method. Beyond FSD50K, our method establishes a scalable paradigm for curating open source audio corpora.