This is the recipe of Japanese female single speaker TTS model with J-KAC dataset.

Before running the recipe, please download from https://sites.google.com/site/shinnosuketakamichi/research-topics/j-kac_corpus. Then, edit 'JKAC' in db.sh and locate unzipped dataset as follows:

$ vim db.sh
JKAC=/path/to/J-KAC

$ tree -L 1 /path/to/J-KAC
/path/to/J-KAC
├── pdf
├── readme.md
├── readme.pdf
├── txt
└── wav

3 directories, 2 files

See the following pages for the usage:

• How to run the recipe
• How to train FastSpeech
• How to train FastSpeech2
• How to train VITS
• How to train joint text2wav

See the following pages before asking the question:

• ESPnet2 Tutorial
• ESPnet2 TTS FAQ