declare-lab/Multimodal-Infomax

Question about the feature extractor

lemonsweetie opened this issue · 2 comments

lemonsweetie commented

Hi, I am confused about the feature extractor in your paper. As I know, COVAREP and P2FA are both feature extractor for acoustic, but you use them for visual and acoustic.

Clement25 commented

Yes, that should be a typo and we are sorry for that. Visual features are extracted with Facet as mentioned in https://arxiv.org/pdf/1906.00295.pdf

lemonsweetie commented

Thanks a lot!