AudioLDM-training-finetuning icon indicating copy to clipboard operation
AudioLDM-training-finetuning copied to clipboard

Published 20 hours ago • verified publisher icon haoheliu

Generating longer audio sequences.

Open Praniyendeev opened this issue 1 year ago • 0 comments

Hi! I was wondering how we could generate a longer audio sequence with the model trained to generate 10 second audio clips? Isnt the Unet architecture fixed and hence the output dimensions the same? the diffusers pipeline seems to do be changing the Unet dimensions,but then dont we need to train it again?

Thank you for your patience, Pranav

Praniyendeev avatar Jan 16 '24 18:01 Praniyendeev