Awesome
Music Genre Classification with Reprogramming
<img src="https://github.com/biboamy/music-repro/blob/main/music-repro.png" width="300">- Low-Resource Music Genre Classification with Advanced Neural Model Reprogramming.
- Yun-Ning Hung, Chao-Han Huck Yang, Pin-Yu Chen, and Alexander Lerch
Codebase
-
Download GTZAN dataset: https://www.kaggle.com/datasets/andradaolteanu/gtzan-dataset-music-genre-classification?resource=download
-
unzip and move file:
unzip archive.zip
mv Data/genres_original/* music-repro/data/
- Install Dependencies
pip3 install -r requirement.txt
- Pull pre-trained models
git lfs fetch --all
- run experiment (skip to "7" for inference only)
python3 main.py --reprog_front uni_noise
python3 main.py --reprog_front condi
python3 main.py --reprog_front skip
- Visit "demo.ipynb" for inference only demo
AST Reference
@inproceedings{gong21b_interspeech,
author={Yuan Gong and Yu-An Chung and James Glass},
title={{AST: Audio Spectrogram Transformer}},
year=2021,
booktitle={Proc. Interspeech 2021},
pages={571--575},
doi={10.21437/Interspeech.2021-698}
}
The ast code used in this repo comes from the original repo
Citing Music Reprogramming
@inproceedings{hung2023low,
title={Low-Resource Music Genre Classification with Cross-Modal Neural Model Reprogramming},
author={Hung, Yun-Ning and Yang, Chao-Han Huck and Chen, Pin-Yu and Lerch, Alexander},
booktitle={Proc. of ICASSP 2023},
pages={1--5},
year={2023},
organization={IEEE}
}
Bug Fix
If you encounter the following errors "batch response: This repository is over its data quota. Account responsible for LFS...", Please download the model from here Google Drive