Guan-Ting/

StyleSpeech-MelGAN-vocoder-16kHz

The MelGAN vocoder for StyleSpeech

About StyleSpeech

• StyleSpeech or Meta-StyleSpeech is a model for Multi-Speaker Adaptive Text-to-Speech Generation
• The StyleSpeech model can be trained by official implementation (https://github.com/KevinMIN95/StyleSpeech).

About MelGAN vocoder

• This MelGAN vocoder is used to transform the mel-spectrogram back to the waveform.
• StyleSpeech is based on 16k Hz sampling rate, and there is no available 16k Hz multi-speaker vocoder.
• Thus I train this vocoder from scratch using Libri-TTS train-100 hour dataset. The training pipeline is the same as the official MelGAN (https://github.com/descriptinc/melgan-neurips).
• The synthesized sounds are close to the official demo with good quality.

Usage

• Please follow the official MelGAN (https://github.com/descriptinc/melgan-neurips) to load pre-trained checkpoint and convert your mel-spectrogram back to the waveform.

Training Details

• GPU: RTX 2080Ti
• Training epoch: 3000