Replies: 1 comment 1 reply
-
|
Also I see Generating audio using F5-TTS in 2 batches. Can we use more batches at once.? |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
I am running on NVIDIA 3060. When producing the audio file it uses between 7-10% of the GPU.
Any 'magic' switches or code changes to speed the process. I changed nfe_step = 16 and this was faster but takes about 8 secs to produce an 8 sec audio.
Using threads ???
Does it re-use the sample audio on each subsequent synthesis ( could it keep this in memory )
.
Any help appreciated.
Beta Was this translation helpful? Give feedback.
All reactions