【KEIRO AI】IDSMILE (アイディスマイル)【Japanese Diff-SVC cover】 by keirokeer_2 published on 2022-12-15T15:31:43Z hi! i trained KEIRO for new Diff-SVC vocoder that support 44.1 kHz model! now he has about 6h20m of data and trained for 122k steps (i need to figure out how to make it better in the steps like 150k+). he is still absolutely supports singing and speaking on English, Russian, Japanese and Chinese! ・original: IDSMILE (アイディスマイル) by toa (とあ) ・vocal: KEIRO AI (Diff-SVC 44.1 kHz) ・vocal reference: CHR0MA (thank you!!) ・ust for harms: Mimisan15, rendered in ACE studio (Que He) ・illust by AI lol listen on YouTube: https://youtu.be/GhmXcbpBuYU ・about me / contacts: https://keirokeer.carrd.co/ midi, ust, tuning, mixing commissions: https://keirokeercom.carrd.co/ Comment by Navo I'm coming back to this song. I wonder If I can attain a similar quality using my own voice and training model with the open source software in the internet? I would love to incorporate this into vocal chops for my personal music! :D 2023-07-26T01:26:25Z Comment by Navo MIND BLOWN!!!!! WHA! This is soooo good. so realistic. 2023-03-09T06:08:14Z Comment by CHR0MA ^^ 2023-01-06T12:33:38Z Comment by Nobooruさん SOUNDS SO ??? OMG??? IT'S SO GOOD 2022-12-15T20:52:50Z