fcaspe / fmtransfer

FM Tone Transfer with Envelope Learning

FM Tone Transfer with Envelope Learning

paper - website

Franco Caspe - Andrew McPherson - Mark Sandler

This is the official implementation of the FM Tone Transfer with Envelope Learning paper, accepted to Audio Mostly 2023.

Install

To install with development dependencies:

$ pip install -e ".[dev]"

Install pre-commit hooks if developing and contributing:

$ pre-commit install

Run

Code in this repo is accessed through the PyTorch Lightning CLI, which is available through the fmtransfer console script. To see help:

$ fmtransfer --help

To run an experiment, pass the appropriate config file to the fit subcommand. For example:

$ fmtransfer fit -c cfg/paper_runs.yaml

To replicate the paper's results, please run:

$ source schedule/test/paper_runs.sh

Citation

If you find this work useful, please consider citing us:

@article{caspe2023envelopelearning,
    title={{FM Tone Transfer with Envelope Learning}},
    author={Caspe, Franco and McPherson, Andrew and Sandler, Mark},
    journal={Proceedings of Audio Mostly 2023},
    year={2023}
}