cvlab-kaist / GaussianTalker

Official implementation of “GaussianTalker: Real-Time High-Fidelity Talking Head Synthesis with Audio-Driven 3D Gaussian Splatting” by Kyusun Cho, Joungbin Lee, Heeji Yoon, Yeobin Hong, Jaehoon Ko, Sangjun Ahn and Seungryong Kim
Other
300 stars 36 forks source link

deepspeech cost too much GPU #53

Open fransonwu opened 2 months ago

fransonwu commented 2 months ago

I have made it, realtime. But deepspeech cost too much GPU, about 24G. So we have to use two GPUs for it. I think the audio features extract process should not be so complicated. Any idea to make it simple and less GPU?