Anonymous 02/05/2023 (Sun) 17:32 No.36782 del
>>36780
>>36781
I think the problem is the ai works better with some source audio than others, and when it struggles it defaults to its pre-trained voices instead of the one you're trying to train it on