UnivNet

UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation.

Training

python train.py --config config_c32.json

Citation

@misc{seo2021controlling,
      title={Controlling Neural Networks with Rule Representations}, 
      author={Sungyong Seo and Sercan O. Arik and Jinsung Yoon and Xiang Zhang and Kihyuk Sohn and Tomas Pfister},
      year={2021},
      eprint={2106.07804},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

Note

• For more complete and end to end Voice cloning or Text to Speech (TTS) toolbox 🧰 please visit Deepsync Technologies.

References:

• Hi-Fi-GAN repo
• LVCNet repo