Trouble with outputs preserving information
#3
by
sdooman
- opened
Hi there, first let me thank you for this model and the great demo/blogpost!
When I use the model with my own .wav files as the input, the output .wav lose a lot of words / information. You mention this in the blogpost, but I'm finding that this is very frequent, enough that it makes the model unusable for my purposes.
I've observed other VC models (this one, which is more specific to singing) that don't seem to have this problem. Is this something inherent to the model, or the training set?