run test_all.py, use audio, get the image

Hangz-nju-cuhk / Talking-Face-Generation-DAVS

Code for Talking Face Generation by Adversarially Disentangled Audio-Visual Representation (AAAI 2019)

MIT License

817 stars 173 forks source link

Closed duguiming111 closed 4 years ago

duguiming111 commented 4 years ago

leezythu commented 4 years ago

@duguiming111 I have encountered the same problem, could you please tell me how to handle it?

IQ17 commented 4 years ago

Hi, same here. Any solution, please?

leezythu commented 4 years ago

我的情况是：如果用作者给的audio，可以产生正常的图片；但是如果换成自己的audio，就会产生完全空白的图片。一个tricky的办法是将自己的audio和作者的audio用软件合并成一个audio，其中作者的audio放在前半部分。我尝试过是可行的。