ashawkey / RAD-NeRF

Real-time Neural Radiance Talking Portrait Synthesis via Audio-spatial Decomposition
MIT License
878 stars 153 forks source link

Audio FPS? #32

Closed Gpwner closed 1 year ago

Gpwner commented 1 year ago

Great Job!

这里的Audio FPS是指训练视频的FPS吗? https://github.com/ashawkey/RAD-NeRF/blob/main/nerf/asr.py#L401 我看您在调用asr.py的时候,没有传入FPS,而代码里默认的FPS=50: https://github.com/ashawkey/RAD-NeRF/blob/main/nerf/asr.py#L44 image

ashawkey commented 1 year ago

@Gpwner No, the video is assumed to be 25FPS, and we always use 2 audio frames (chunks) per video frame, and that's the meaning for 50.

Gpwner commented 1 year ago

thanks for reply