I get inconsistent gibberish with fr_FR mls model (with 125 voices)
#67
by YagCed - opened
Hi,
Pretty much all is in the title^^
Everytime i try to get a wav file from the model (whatever the sid) i get a different gibberish every run.
Is this model still supported ?
Thanks
These models don't work well with short phrases. They were an experiment based on audiobook training data, and I'd say the experiment failed ☹️
I see, too bad, a choice of 125 different voices was very interesting to me, and, listening the demos, they didn't sound bad.
Anyway, thanks for taking the time to respond and thanks for the info :-)