I'm trying to train a model for singing and I was wondering if it's better to just train it on the character talking or if I should also include singing in the samples.
#Training a model for singing. Should I train talking, or talking and singing?
high ocean
sharp jacinth
if its specifically for singing, its better to use a singing dataset so it gets how the original voice sounds like when singing
since some people have different voices when talking and singing
high ocean
So I should do solely singing?
still spire
i mean hybrid models are possible, i did them before and they turn out fine
high ocean
Got it, thanks for the help :)
smoky cedar
It works with both: talking and singing.
earnest flume
You can make a talking model and it could still sing (although very limited in range).
It's good to have that range, so you can add a little bit of talking