https://uberduck.ai/ logo
Join Discord
Powered by
# machine-learning
  • o

    OctolingVladTTS

    04/27/2021, 10:16 PM
    but I'm lazy looking for them If you want, find the dialogues and send them to me in wav format to transcribe them
  • p

    pi

    04/27/2021, 10:16 PM
    then please stop asking why it doesn't work with less 100 seconds of voice samples
  • p

    pi

    04/27/2021, 10:17 PM
    but you want the voice !!
  • o

    OctolingVladTTS

    04/27/2021, 10:17 PM
    it's okay
  • n

    n1cte

    04/27/2021, 10:18 PM
    Thanks for sharing the website
  • o

    OctolingVladTTS

    04/27/2021, 10:50 PM
    If you want, look for the dialogues that reba buhr performed to transcribe them
  • o

    OctolingVladTTS

    04/27/2021, 11:00 PM
    I hope they find it
  • k

    Kosmikist

    04/27/2021, 11:56 PM
    Im not sure if its just the voice clips I’m using but the Lemon Demon model Ive been working on isnt making that much of comprehensible words. Anyone have an idea on what the problem could be? Heres a link of the model I’ve trained so far: https://drive.google.com/file/d/1eWYKx2WVH9-1p3thoFnHrjtFk5giq0lu/view?usp=drivesdk
  • p

    pi

    04/28/2021, 12:12 AM
    post your dataset
  • p

    pi

    04/28/2021, 12:16 AM
    also: they are a music group; tacotron is not capable making music; is that what you're trying within tacotrons capabilities?
  • m

    mega b

    04/28/2021, 12:42 AM
    1. more training data 2. sounds like speeding bullets
  • m

    mega b

    04/28/2021, 12:42 AM
    3. turn on raw_input
  • o

    OctolingVladTTS

    04/28/2021, 1:11 AM
    i'm working with luigi's voice
  • f

    F.B

    04/28/2021, 1:15 AM
    Good luck with that mate
  • o

    OctolingVladTTS

    04/28/2021, 1:16 AM
    I put 52 voice lines to see if the voice comes out right
  • p

    PUMPKINEATER

    04/28/2021, 1:16 AM
    Good luck.
  • o

    OctolingVladTTS

    04/28/2021, 1:16 AM
    Yep
  • o

    OctolingVladTTS

    04/28/2021, 1:17 AM
    I hope it work me
  • p

    pi

    04/28/2021, 1:17 AM
    your samples are still very short
  • p

    pi

    04/28/2021, 1:17 AM
    better get more
  • p

    pi

    04/28/2021, 1:18 AM
    it's about total speech time not about sample (I told you before)
  • d

    Dimitri Rascalov

    04/28/2021, 1:34 AM
    and i'm stuck here, but anyways good luck
  • k

    Kosmikist

    04/28/2021, 1:37 AM
    The source audio is singing though its the character from friday night funkin, not the music group
  • d

    Dimitri Rascalov

    04/28/2021, 1:38 AM
    forget it, I fixed it
  • d

    Dimitri Rascalov

    04/28/2021, 1:38 AM
    it was just a little mistake of mine
  • p

    pi

    04/28/2021, 1:39 AM
    it's really not enough data
  • p

    pi

    04/28/2021, 1:40 AM
    and also not the kind of data tacotron does well on
  • k

    Kosmikist

    04/28/2021, 1:40 AM
    Ah alright, well it was worth a try
  • p

    pi

    04/28/2021, 1:41 AM
    keep it somewhere and hope that eventually an architecture comes around that can deal with this data
  • p

    pi

    04/28/2021, 1:42 AM
    🙂
1...171819...1068Latest