NNSVS recipe of Yuuri Natsume's singing voice database (51 songs). Almost all codes are derived from kiritan_singing.
The recipe of Yuuri Natsume's singing voice database is merged into NNSVS repository on 04 Nov 2020. Please use the official one.
This repository is maintained only for the experimental purpose.
- nnsvs
- pysinsy
- nnmnkwii
- librosa
- soundfile
- scipy
- numpy
- tqdm
- jaconv
- pyyaml
Due to the licensing issue, this recipe does not include data nor helper scripts for downloading automatically. First of all, you need to get Natsume_Singing_DB.zip from 夏目悠李/男声歌声データベース配布、始めました!【2020/10/10更新】 (the terms of service are written in Japanese). Next, clone this repository and change db_root
in 00-svs-world/config.yaml
according to your environment. Then move to 00-svs-world
directory and run:
run.sh --stage 0 --stop-stage 6
The directory structure made by this recipe is the same as kiritan_singing does.
- Jupyter Notebook(Using the official recipe) (Google Colaboratory, comments are written in Japanese)
- Jupyter Notebook(Older version)
- 夏目悠李/男声歌声データベース配布、始めました!【2020/10/10更新】: https://amanokei.hatenablog.com/entry/2020/04/30/230003