Ryuichi Yamamoto

Sorry that's my bad. I didn't include any detailed information in the description. Some notes: - `baseline`: a baseline ResSkipF0FFConvLSTM model - `gan`: my attempt to integrate GAN for training...

A good news: I've done an initial cut for MS post-filter and here is the spectrogram example: From top to bottom: `gan`, `gan` with MS post-filter, `neutrino`  Findings so...

 Top: NNSVS (w/ GAN-based post-filter) Bottom: Neutrino My bad; previous spectrogram visualization was wrong. I was assuming that neutrino uses the same mgc as ours, but it turned...

報告ありがとうございます。https://github.com/r9y9/pyopenjtalk/issues/34 こちらと同一だと思いますが、私の環境で再現できるかどうか確認してみます。

Google colab、私のmacでは再現しませんでした。Googleで検索すると似たような問題 https://github.com/pypa/manylinux/issues/178#issuecomment-528520527 が見つかりますが、pyopenjtalkやttslearnの問題ではなく、cmakeに関連する問題だと思います。コマンドラインから cmake を実行して、同様のエラーが出ないか、cmakeのバージョンを変えることで解決しないか、複数のcmakeがインストールされていて予期せぬ動作を及ぼしていないかなど、調べていただけますでしょうか

To be honest, I am also waiting for contribution. I may try making recipes at some point but not sure if I can do it since I don't speak Chinese...

I guess converting textgrid files into HTK labels is not that difficult. That said, I plan to seek other input types rather than HTK labels to implement opencpop recipes easily.

For the record, the amount of data: - 5.21 hours - 4.35 hours (excluding SP/AP segments) Noticed that a fair amount of data (approx. 16%) is considered non-voiced sounds.

List of utterances that have long (>5 sec.) silences: - 2020000775 - 2020000795 - 2048001806 - 2048001826 - 2087003218 Not sure why these long silences are included.

Uploaded samples to https://r9y9.github.io/projects/nnsvs/. Will make a PR for the recipes after #175 gets merged.