# machine-learning
- c
CyberSage87
07/05/2021, 10:00 PMThat sounds amazing! How many wavs were put into that? - u
_n
07/05/2021, 10:03 PM371 clips, 21:23 total length - u
user
07/05/2021, 10:03 PMHow do you guys do it? - u
_n
07/05/2021, 10:03 PMthe dataset isnt very good though, i have a lot of shouty, stuttery and emotive lines - u
user
07/05/2021, 10:04 PMHow do you guys use the character voices and alter them? - u
user
07/05/2021, 10:04 PMWith all the dataset and stuff? - u
_n
07/05/2021, 10:05 PMi dont remember if i uploaded it or not, you can have a look though it - h
Hoodied
07/05/2021, 10:09 PMWhat batch did u put it as - u
_n
07/05/2021, 10:12 PMAI TTS synthesis uses machine learning to create a model. the model is trained using a dataset from audio files that have been transcribed. after training for enough epochs, you can use the model to synthesise new audio from text. if you want to get started, there's a more detailed guide at https://awesomeface349.gitbook.io/uberduck-written-tutorial/ - a
Awesome Face (GTX 1660 Ti)
07/05/2021, 10:14 PMstill waiting on update for loading exsisting transcription file - u
_n
07/05/2021, 10:17 PMyeah sorry, ive been busy with university and graduation and life stuff lol - a
Awesome Face (GTX 1660 Ti)
07/05/2021, 10:17 PMah, that's fine - a
Awesome Face (GTX 1660 Ti)
07/05/2021, 10:17 PMyeah... everyone has lifes - s
scruffygamer (SYNERGY X CREATOR)
07/05/2021, 10:32 PMWthhhhhh - s
scruffygamer (SYNERGY X CREATOR)
07/05/2021, 10:32 PMIt started like this lool - c
Couch
07/05/2021, 10:33 PMdelete hifigan folder in output directory - s
scruffygamer (SYNERGY X CREATOR)
07/05/2021, 10:33 PMWhere - s
scruffygamer (SYNERGY X CREATOR)
07/05/2021, 10:33 PMI'm finetuning - s
scruffygamer (SYNERGY X CREATOR)
07/05/2021, 10:38 PMI fixed it. - u
user
07/06/2021, 1:20 AMi managed to get up to 10 minutes and 173 wav files so i guess i'll see how it much better it is this time around with about 4 extra minutes of audio compared to last night - c
Cioss
07/06/2021, 11:33 AMy'all could probably add bennett foddy - c
Cioss
07/06/2021, 11:33 AMiirc he isn't a voice yet - h
Hoodied
07/06/2021, 3:12 PMI think this sounds really good so far! - s
SidPlays_144p
07/06/2021, 4:03 PMthis is some funky new Taco Bell ad - s
Scramble
07/06/2021, 5:13 PM#842148452464853002, #842056330835656745 /voicesuggest - v
v0id
07/06/2021, 5:32 PMgameing - a
ABold
07/06/2021, 5:41 PMTakco - m
mega b
07/06/2021, 7:25 PMGood stuff, link the colab here. - b
BigDRᗩCO$O
07/06/2021, 7:37 PMI'm thinking about making a playboi carti model. so what carti do yall prefer? Baby voice or old Carti @everyone - a
Aniball
07/06/2021, 7:52 PMThe images were made using a bot in EleutherAI's discord server