Taming Data and Transformers for Audio Generation

TL;DR

This paper introduces AutoReCap-XL, AutoCap, and GenAu, a large-scale dataset, high-quality captioning model, and scalable transformer architecture for ambient audio generation, significantly improving quality and scalability.

Contribution

It presents a comprehensive approach combining data collection, captioning, and scalable modeling to advance ambient audio generation.

Findings

AutoReCap-XL dataset with over 47 million clips

AutoCap achieves CIDEr score of 83.2, 3.2% better than previous models

GenAu improves FAD by 4.7%, IS by 11.1%, and CLAP score by 13.5%

Abstract

The scalability of ambient sound generators is hindered by data scarcity, insufficient caption quality, and limited scalability in model architecture. This work addresses these challenges by advancing both data and model scaling. First, we propose an efficient and scalable dataset collection pipeline tailored for ambient audio generation, resulting in AutoReCap-XL, the largest ambient audio-text dataset with over 47 million clips. To provide high-quality textual annotations, we propose AutoCap, a high-quality automatic audio captioning model. By adopting a Q-Former module and leveraging audio metadata, AutoCap substantially enhances caption quality, reaching a CIDEr score of $83.2$, a $3.2\%$ improvement over previous captioning models. Finally, we propose GenAu, a scalable transformer-based audio generation architecture that we scale up to 1.25B parameters. We demonstrate its benefits from data scaling with synthetic captions as well as model size scaling. When compared to baseline audio generators trained at similar size and data scale, GenAu obtains significant improvements of $4.7\%$ in FAD score, $11.1\%$ in IS, and $13.5\%$ in CLAP score. Our code, model checkpoints, and dataset are publicly available.