Is it possible to improve the accuracy of the voice clone? Maybe retraining the code?
#7
by
Housebotz
- opened
I make for sure that the audio is as clean as possible, yet the voice clone isn't as accurate at copying the voice of the original sample.
it doesn't copy it, it mimics it the best it can. it is a quick train model. i get amazing results with my 15 minute mark rober model (it is a bit fast) to the point where it is instantly recognisable as mark rober. of corse, it isn't perfect and wont trick anyone