# machine-learning
- u
user
04/05/2022, 2:29 AMi perceive reference mode as a mode to have characters speak with more emotion and better inflection, whoops, forgot it was actually about singing - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:30 AMwell that was the original point of talknet i think - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:30 AMwhen like nvidia made it - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:30 AMbut cookie made it more of a singing archetype - u
user
04/05/2022, 2:31 AMNvidia's made great strides with filtering out the sound of a keyboard when someone speaks, but doesn't bother to improve Talknet, that is...if Nvidia DID make it - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:31 AMin fact pitch correction was something of cookie’s doing - u
user
04/05/2022, 2:31 AMi do not know this "cookie" person you speak of - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:31 AMsomeone from pony preservation project i believe - u
user
04/05/2022, 2:33 AMrrrright then - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:33 AMoh hang on - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:34 AMi think it was sortanon that did the talknet stuff - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:34 AMcookie did the tacotron stuff - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:34 AMi need to check the notebooks - u
user
04/05/2022, 2:34 AMwhenever i think of the word "pony" all i can think of is ||15, because of his profile picture and his interests and how the MLP ("multi-layer perceptions") community contributed to his website|| - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:36 AMjust checked, talknet notebook only credits PPP but tacotron is indeed credited to cookie - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:36 AMstill pretty sure it was sortanon that did the talknet stuff - u
user
04/05/2022, 2:47 AMhey, uh, off-topic-yet-important, would you happen to know what sam's time-zone is - u
user
04/05/2022, 2:47 AMit would explain his/her inactivity - u
(Dawn) Will Draw Fictional Women
04/05/2022, 2:47 AMnah - u
{K EY1} (Kei)
04/05/2022, 3:03 AMI mean If you want the spamton model back You could make your own 💀 - u
user
04/05/2022, 3:05 AMoh, that's a GREAT idea, why didn't i think of that - u
user
04/05/2022, 3:05 AMi already submitted a model with voice samples and a transcription TXT file hours ago - u
user
04/05/2022, 3:06 AMand i'm currently trying to "train" his voice using some real-time voice cloning as we speak - u
{K EY1} (Kei)
04/05/2022, 3:06 AMCool uh Sorry? The sarcastic response was unnecessary - u
user
04/05/2022, 3:07 AMi'm just a little miffed about the only remotely-accurate sounding Spamton AI voice being removed is all - u
user
04/05/2022, 3:09 AMbut of course, that doesn't justify my remark, however it does explain why i responded with it - h
hecko
04/05/2022, 5:50 AMi wonder if maybe training it in a multispeaker along with e.g. transformers voices would help - a
Alphabet
04/05/2022, 5:57 AMwow, now that is super interesting that audo.ai was able to remove background chatter from a restaurant. That basically opens up most tiktok personalities now if that site is able to rip vocals that well. Interesting - a
Alphabet
04/05/2022, 6:37 AMok, so audo.ai is able to remove all non-vocal and random background noises. Footsteps? gone. Random people in back talking? gone . Eating a meal and there's utensils or chewing sounds? gone. - a
Alphabet
04/05/2022, 6:37 AMthat's amazing