Raw LLM Responses

Basically people are using so-vits-svc to train ai inference models on studio acapellas or extracted acapellas of famous singers that can make one voice tone and character sound like another — there’s a cloud code notebook people are sharing that allows you to upload your own dry vocal and the output is the inference of your vocal sounding like the pretrained so-vits-svc inference model you choose (drake, kanye etc). Most people are extracting acapellas and making songs sound like drake with a pre trained model but the ghostwriter guy who did the drake / weeknd “ai song” most likely wrote and sang the vocals himself trying to emulate each artist and then used the so-vits-svc models to make it sound very realistic. He may have used chatgpt for the lyrics but most likely altered some or wrote it himself.