English demo?

[impactcolor]

Thank you for sharing the colab notebook. Is there an example of english demo? In the current colab I have done the following:

I changed pretrained acoustic model to this: config_path = "downloads/zh/muskit/exp/svs_with_lid/config.yaml" model_path = "downloads/zh/muskit/exp/svs_with_lid/119epoch.pth"

Which is from: Multilingual_four

Then I changed pretrained vcoder to this: vocoder_config = "downloads/zh/pwg/config.yml" vocoder_checkpoint = "downloads/zh/pwg/checkpoint-1000000steps.pkl" print("successfully finished download vocoder")

Which is one of the models from results.
The final blocks gives me the error: 31 batch = sing_generation.preprocess_fn(str(row["id"]), info, 1.0) 32 batch = tensorify(batch) ---> 33 singing, _, _, _, _, _, _ = sing_generation(**batch)

I'm obviously doing something wrong. Can you give me any insight into how to do english models can be loaded?

[ftshijt]

Hi, thanks for your interest. Most of our contributors are in winter vacation due to the Chinese new year. Will back to you as soon as we are back. Before that, could you share your notebook with logs for us to check the issues?

[ftshijt]

Btw, FYI, we have merged most of the functionality of Muskits to ESPnet, where we gonna move our focus to there in the near future. We already have the SOTA VISinger in ESPnet-Muskits now.

[impactcolor]

Hi thanks for your response. Is there a notebook for ESPnet-Muskits? @ftshijt