SoloAudio: Target Sound Extraction with Language-oriented Audio Diffusion Transformer

TL;DR

SoloAudio is a new diffusion-based model for target sound extraction that uses a Transformer on latent features and incorporates language-oriented features, achieving state-of-the-art results and strong generalization to unseen sounds.

Contribution

It introduces a Transformer-based latent diffusion model with language-oriented features for target sound extraction, enhancing generalization and zero-shot capabilities.

Findings

Achieves state-of-the-art results on FSD Kaggle and AudioSet datasets.

Demonstrates strong zero-shot and few-shot sound extraction performance.

Shows improved generalization to out-of-domain and unseen sound events.

Abstract

In this paper, we introduce SoloAudio, a novel diffusion-based generative model for target sound extraction (TSE). Our approach trains latent diffusion models on audio, replacing the previous U-Net backbone with a skip-connected Transformer that operates on latent features. SoloAudio supports both audio-oriented and language-oriented TSE by utilizing a CLAP model as the feature extractor for target sounds. Furthermore, SoloAudio leverages synthetic audio generated by state-of-the-art text-to-audio models for training, demonstrating strong generalization to out-of-domain data and unseen sound events. We evaluate this approach on the FSD Kaggle 2018 mixture dataset and real data from AudioSet, where SoloAudio achieves the state-of-the-art results on both in-domain and out-of-domain data, and exhibits impressive zero-shot and few-shot capabilities. Source code and demos are released.