this is a fuse model to make midi+lyric to a fuse embedding that can generate this music's cover at this project, I use ...
first use source ./modelEnv/env388/bin/activate into virtual envirment
use python3 generate_image.py
and input artist and song name
it will auto download song's lyrics and music
and generate its cover image
the similarity of fuse embedding and original cover
the loss of decoder
![]()
- dataset
- bert
- diffusion model
- microsoft musicBert
- YourMT3 this is a model for transform wav to midi
- LP-MusicCaps
- Qiu_Image_Generation_Associated_CVPR_2018_paper
- Performance-MIDI to Score
- DALI
- image output size is only 64x64
- decoder overfitting
- 葉梅珍 教授
- 王鈞右 教授
- 王兆偉
- 鍾詠傑
- 吳文元