Audio-FLAN: A Preliminary Release

TL;DR

Audio-FLAN introduces a large-scale, diverse instruction-tuning dataset for audio tasks, enabling the development of unified models capable of understanding and generating speech, music, and sounds in a zero-shot setting.

Contribution

It provides the first comprehensive dataset covering both audio understanding and generation tasks, facilitating unified audio-language modeling.

Findings

Dataset includes over 100 million instances across 80 tasks.

Enables zero-shot learning for diverse audio tasks.

Supports development of unified audio-language models.

Abstract

Recent advancements in audio tokenization have significantly enhanced the integration of audio capabilities into large language models (LLMs). However, audio understanding and generation are often treated as distinct tasks, hindering the development of truly unified audio-language models. While instruction tuning has demonstrated remarkable success in improving generalization and zero-shot learning across text and vision, its application to audio remains largely unexplored. A major obstacle is the lack of comprehensive datasets that unify audio understanding and generation. To address this, we introduce Audio-FLAN, a large-scale instruction-tuning dataset covering 80 diverse tasks across speech, music, and sound domains, with over 100 million instances. Audio-FLAN lays the foundation for unified audio-language models that can seamlessly handle both understanding (e.g., transcription, comprehension) and generation (e.g., speech, music, sound) tasks across a wide range of audio domains in a zero-shot manner. The Audio-FLAN dataset is available on HuggingFace and GitHub and will be continuously updated.