Show dataset

[lovemino]

Thanks a lot for open-sourcing your code and for your awesome contribution!
I’m looking to build on your project for some further exploration, and I was hoping you might be willing to share the data preprocessing scripts or the test set for the Show dataset?
Thanks again for your kindness!

[Xiangyue-Zhang]

Thanks a lot for open-sourcing your code and for your awesome contribution! I’m looking to build on your project for some further exploration, and I was hoping you might be willing to share the data preprocessing scripts or the test set for the Show dataset? Thanks again for your kindness!

Thank you very much for your interest in our work and for your kind words about SemTalk.

Unfortunately, the SHOW dataset cannot be shared at the moment. I have already left my previous company, and I no longer have the permission or access required to release the dataset or the related preprocessing scripts.

As an alternative, we suggest that you consider using the BEAT2 motion capture dataset. In particular, you may try validating your method under the Speaker-2 and All-Speaker settings, which we believe can serve as a reasonable testbed for similar evaluations.

Thank you again for your interest in SemTalk, and we appreciate your understanding 🙏