MusicEval (AISHELL-7A)
Generative music evaluation dataset with expert ratings for text-to-music samples
Duration
16.62 hours
Languages
1
Sample Rate
16 kHz
Published
2025-01
Description
1MusicEval is a generative music assessment dataset
2The total duration is 16.62 hours, containing 2748 mono music clips
3The clips were generated by 21 systems covering 31 models in response to 261 text prompts
4Includes 13740 ratings from 14 music experts
5Among the 261 text prompts, 80 were manually written, 20 came from MusicCaps, and the remaining 161 correspond to demo-only systems
6All music clips are 16 kHz mono audio
7Each music clip was evaluated by 5 music experts on overall musical impression and alignment with the text prompt
Language Details
| Language | Duration |
|---|---|
| None | None |
Publisher
License & Commercial Use
License
CC BY-NC 4.0
Commercial Use
Non-commercial only