< Back to IRCAM Forum

Voice model training

Hi, I’m about to train a model on acapella vocalizations, have you got any tips on specific code setups in rave to optimize the output ?

Hello @scorrazzi ! I would advise you to use either a v1 or v2 configuration. You can try the --config causal additional configuration, if you really need fast timbre transfer ; at the cost of a lower reconstruction quality. Thanks for your interest!

1 Like