Open pmcanneny opened 1 year ago
I get a performance boost when using GPU, but inference is still much slower than I'd expect.
For context, inference can take up to 8 seconds, but I can generate 512x512 stable diffusion images in under 1 second on my current hardware setup.
Is this issue still accurate? https://github.com/CorentinJ/Real-Time-Voice-Cloning/issues/1013
Is there an easy way for me swap in a different vocoder to test for speed/quality increases?
I get a performance boost when using GPU, but inference is still much slower than I'd expect.
For context, inference can take up to 8 seconds, but I can generate 512x512 stable diffusion images in under 1 second on my current hardware setup.
Is this issue still accurate? https://github.com/CorentinJ/Real-Time-Voice-Cloning/issues/1013
Is there an easy way for me swap in a different vocoder to test for speed/quality increases?