I'm a comedy genius

I think I've mentioned that I unironically use DiffSinger training as way for PC to act as a space heater for my office (in this case it's the room labeled as the "downstairs bedroom") and like.

I recorded with Mariya (AERIS' VP) today for DiffSinger! We had a very productive recording session, recorded an hour of a base voice, plus power and soft vocal modes, talking and rap data; as well as Russian and Ukrainian!

I'm back out here making sketchy recording booths in Airbnbs for recording DiffSinger

What if I tried training DiffSinger on my Asus ROG Ally X with an eGPU? 😳

Gonna be trying something interesting with Merry's DiffSinger today, labeling some of her UTAU pitches for legacy voice colours with genon2nnsvs.

When I was in France, I got 1 person to record DiffSinger for me, it's not a lot of data (10 mins of singing and some speaking); should be fine with the mega model. Anyway I finished labeling the singing today, might also label the random speaking but I'm lazy (and French people talk fast)

Everyone is coming up with such creative names for their AI Bank's vocal modes and then there's me who's like. Ah yes Power, Soft, Cute, Opera that works.

Ayy, so I'm still on Twitter (same @; ViolinBaguette) but I'll be crossposting my posts here too