by hujinsen

Fully reproduce the paper of StarGAN-VC. Stable training and Better audio quality .

162 Stars 36 Forks Last release: Not found 10 Commits 0 Releases

Available items

No Items, yet!

The developer of this repository has not created any items for sale yet. Need a bug fixed? Help with integration? A different license? Create a request here:


This is a pytorch implementation of the paper: StarGAN-VC: Non-parallel many-to-many voice conversion with star generative adversarial networks.

The converted voice examples are in samples and results_2019-06-10 directory


  • Python 3.6+
  • pytorch 1.0
  • librosa
  • pyworld
  • tensorboardX
  • scikit-learn


Download dataset

Download the vcc 2016 dataset to the current directory


The downloaded zip files are extracted to

  1. training set: In the paper, the author choose four speakers from
    . So we move the corresponding folder(eg. SF1,SF2,TM1,TM2 ) to
  2. testing set In the paper, the author choose four speakers from
    . So we move the corresponding folder(eg. SF1,SF2,TM1,TM2 ) to

The data directory now looks like this:

├── speakers  (training set)
│   ├── SF1
│   ├── SF2
│   ├── TM1
│   └── TM2
├── speakers_test (testing set)
│   ├── SF1
│   ├── SF2
│   ├── TM1
│   └── TM2
├── vcc2016_training (vcc 2016 training set)
│   ├── ...
├── evaluation_all (vcc 2016 evaluation set, we use it as testing set)
│   ├── ...


Extract features (mcep, f0, ap) from each speech clip. The features are stored as npy files. We also calculate the statistical characteristics for each speaker.


This process may take minutes !




python --mode test --test_iters 200000 --src_speaker TM1 --trg_speaker "['TM1','SF1']"

Network structure


Note: Our implementation follows the original paper’s network structure, while pytorch StarGAN-VC code use StarGAN's network.Both can generate good audio quality.


tensorflow StarGAN-VC code

StarGAN code

CycleGAN-VC code

pytorch-StarGAN-VC code

StarGAN-VC paper

StarGAN paper

CycleGAN paper

Update 2019/06/10

The former implementation's network structure is the network of the original paper, but in order to achieve better conversion result, the following modifications are made in this update: - Modification of classifier without training problem - Update loss function

- Modify the discriminator activation function to tanh

If you feel this repo is good, please star !

Your encouragement is my biggest motivation!

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.