>>79119384It's not feasible if you want to do asmr or close proximity voices. I've been writing jp asmr scripts for a living since 2018, but when I tried making some stuff of my own using AI voice changers- stuff like ear licking, swallowing sounds, and even breathing distorts the output sound so easily and there's also so much artifacting noise. Any minor pitch changes throw it off too. And this was all feeding the AI with pre-recorded stuff, not live encoding.
This was back in late 2023, but I'll let it cook for another 5 years. There needs to be sound engineers and audio specialists giving feedback to these AI devbros. It's just way too far off ideal right now.