(the FULL audio files and jsonl files are still updating)
An Instruction-Tuning Dataset for Unified Audio Understanding and Generation Across Speech, Music, and Sound.
The Audio-FLAN-Dataset has the following directory structure:
Audio-FLAN-Dataset/
├── audio_files/
│ ├── audio/
│ │ └── 177_TAU_Urban_Acoustic_Scenes_2022/
│ │ └── 179_Audioset_for_Audio_Inpainting/
│ │ └── ...
│ ├── music/
│ │ └──… See the full description on the dataset page:
https://huggingface.co/datasets/HKUSTAudio/Audio-FLAN-Dataset.
