Open light1726 opened 4 years ago
By the way, do you cut the input sequence into fixed-length which is the multiple of the downsampling frequency during training? if so how long is the fixed-length?
Thanks. The code is correct. 2 seconds.
I see, thanks for the answer.
Thanks for sharing the code and you did a great job. I noticed that in the paper the downsampling process on the temporal axis is different for the forward sequence and the backward sequence. But it seems that the downsampling operation for the forward sequence in the code follows exactly the process described in the paper for the backward sequence. I'm quite confused because these two processes (what described in the code and in the paper) seems to behave differently for that they encode different contextual information. Since the code is more up-to-date, does the downsampling process in the code is better?
https://github.com/auspicious3000/autovc/blob/2d8a6c8856f59c1c2af5cf8d4143b0a9605fbe0e/model_vc.py#L79