jixinya

> @auspicious3000 Thanks for your suggestion, I have trained the 80 speakers(P225~P304) in VCTK dataset(due to onehot size is 80) on 2080Ti GPU for 2 days, the result becomes better,...

> > Sorry to bother you again, but I am still a little confused about the preprocessing of ATnet and VGnet. I didn't find explicit code for preprocessing the training...

Thanks! There's one more question. Did you use the method in 'Talking-Face-Landmarks-from-Speech' to frontalize the landmarks in ATnet? Cause when I try normLmark() in demo.py to process the data, I...

I haven't met this problem before, but I guess it might comes from the missing part of ears or neck of the edge map (as shown below). 

我这里显示下载没有问题，或许可以再刷新试试看

I have released the training code. More details of DTW can be seen in train/disenetanglement/dtw/MFCC_dtw.py.

Hi, we use a landmark detector to detect the 106 facial key points of each frame. However, we can not provide the detector here due to copyright reasons. You can...

I have released the training code. You can check it in train/disentanglement/code/models_content_cla.py (class Decoder).

The output naming rules of M003 are different from others (M009,M030 etc.) due to the update of MEAD dataset. The old rule follows the order of sentences while the new...

The results don't have obvious eye blinks since we did not intentionally add eye blinks in the video.