Skip to content
TensorFlow implementation of "GANSynth: Adversarial Neural Audio Synthesis"
Python
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
README.md update Apr 12, 2019
dataset.py update May 17, 2019
gan_synth_main.py update May 17, 2019
make_tfrecord.py
metrics.py update May 17, 2019
models.py update May 17, 2019
networks.py update May 17, 2019
ops.py update Apr 16, 2019
pitch_classifier_main.py update May 17, 2019
spectral_ops.py update Apr 20, 2019
utils.py update May 17, 2019

README.md

GANSynth: Adversarial Neural Audio Synthesis
TensorFlow implementation of the ICLR 2019 paper

Original paper

Based on following papers

Dataset

Requirements

  • TensorFlow 1.13.1 with GPU support.

Usage

  • Following the paper, create a new train/valid/test 80/10/10 split from shuffled data, as the original split was divided along instrument type, which isn’t desirable for this task.

  • The original tfrecord is very large, so it takes so long time to shuffle after each epoch.
    For better peformance, make tfrecord which contains the path to waveform.

wget http://download.magenta.tensorflow.org/datasets/nsynth/nsynth-train.jsonwav.tar.gz
wget http://download.magenta.tensorflow.org/datasets/nsynth/nsynth-valid.jsonwav.tar.gz
wget http://download.magenta.tensorflow.org/datasets/nsynth/nsynth-test.jsonwav.tar.gz

tar -xvf nsynth-train.jsonwav.tar.gz
tar -xvf nsynth-valid.jsonwav.tar.gz
tar -xvf nsynth-test.jsonwav.tar.gz

python make_tfrecord.py
python main.py --filenames nsynth_train_examples.tfrecord --train
python main.py --filenames nsynth_test_examples.tfrecord --evaluate
You can’t perform that action at this time.