https://uberduck.ai/ logo
Join Discord
Powered by
# tacotron-2-support
  • r

    Radak

    02/09/2023, 5:14 AM
    It’s the up to date version
  • i

    I Fucking Hate Papa Jons

    02/09/2023, 6:35 AM
    is it possible to train models on higher quality audio?
  • i

    I Fucking Hate Papa Jons

    02/09/2023, 6:35 AM
    like maybe just 44100 hz or in stereo, cause i'm kinda tired of the 22050hz mono limit to be honest but if it isn't possible i'll roll with what we still have i haven't done tts training in a while so idk if there's anything new
  • g

    Gosmokeless28

    02/09/2023, 6:44 AM
    Not really as of now
  • i

    I Fucking Hate Papa Jons

    02/09/2023, 6:45 AM
    oh that sucks
  • u

    (Dawn) Will Draw Fictional Women

    02/09/2023, 6:56 AM
    it’s theoretically possible with tacotron
  • u

    (Dawn) Will Draw Fictional Women

    02/09/2023, 6:57 AM
    but training has proven to be incredibly unstable
  • j

    jalin28

    02/10/2023, 1:53 AM
    I'm getting a error like this where I need to run step 1 like I already did im not sure why I'm getting this error
  • g

    Gosmokeless28

    02/10/2023, 1:56 AM
    That's odd. Something must be corrupt. I don't know why that error is occurring.
  • j

    jalin28

    02/10/2023, 1:57 AM
    Yeah I know right
  • g

    Gosmokeless28

    02/10/2023, 1:58 AM
    I need @Cris140's help again
  • j

    jalin28

    02/10/2023, 1:58 AM
    Ok
  • j

    jalin28

    02/10/2023, 1:59 AM
    Yeah like I did step one and I don't know why that error is occurring
  • c

    Cris140

    02/10/2023, 2:40 AM
    Which notebook are you using?
  • g

    Gosmokeless28

    02/10/2023, 3:30 AM
    Look at the screenshot
  • c

    Cris140

    02/10/2023, 3:42 AM
    Oh, I see
  • c

    Cris140

    02/10/2023, 3:42 AM
    I don't use the Import files option
  • c

    Cris140

    02/10/2023, 3:42 AM
    I can check if there's any problem with it
  • c

    Cris140

    02/10/2023, 3:42 AM
    But if you have less than 150 wavs, it's just better to drag'n'drop on the wavs folder
  • j

    jalin28

    02/10/2023, 10:13 AM
    Oh I see That's why
  • d

    DDark

    02/10/2023, 11:20 PM
    I can't download it frkm github there is no download button
  • d

    DDark

    02/11/2023, 12:24 AM
    I still can't find the download button
  • d

    DDark

    02/11/2023, 12:30 AM
    Hello?
  • t

    TwoTrickOmnic

    02/11/2023, 11:44 AM
    question for a dataset of around 7 minutes, how many epochs should i train to (i am a complete noob)
  • t

    TwoTrickOmnic

    02/11/2023, 11:48 AM
    is 35 good
  • j

    jalin28

    02/11/2023, 3:50 PM
    I have made my first tacotron2 model now
  • g

    Gosmokeless28

    02/11/2023, 9:42 PM
    200 is good
  • g

    Gosmokeless28

    02/11/2023, 9:42 PM
    Have you submitted it to Uberduck?
  • j

    jalin28

    02/11/2023, 9:44 PM
    Yeah it's right here and it got approved but the thing is I won't let me make it primary
  • g

    Gosmokeless28

    02/11/2023, 9:46 PM
    That's because only admins & mods can make voice models primary
1...122123124...158Latest