Amphion

So I wanted some voice conversion to make Fam’s speech less robotic yet my findings resulted in:

• PyWorld: Male+Female voices = Young Boy
• SoftVC VITS Singing Voice Conversion: Outdated
• Amphion: This mixes a lot into one package, making it huge (14.6 Gb just for Vevo)!

Which leans a lot into using Machine Learning and thus, requiring expensive GPUs or you’re stuck with something basic like PyWorld.

I have found an interesting gui and a video showcasing it which helps quite a bit but also shows its limitations.