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

    user

    05/22/2021, 10:41 PM
    you have the model?
  • l

    LindosoLord

    05/22/2021, 10:42 PM
    Yep
  • u

    user

    05/22/2021, 10:42 PM
    can you post it?
  • l

    LindosoLord

    05/22/2021, 10:43 PM
    ok
  • w

    Waltman13 (semi-active)

    05/22/2021, 10:54 PM
    Doing a quick test with the first model by adding 16 additional wav files to it (which are from the actual show and are gonna be used for V2), and so far... yeahh this isn't looking too good..
  • w

    Waltman13 (semi-active)

    05/22/2021, 10:56 PM
    it might be due to the new clips being from PAL rips of the episodes (which are usually at a higher tone), but I'll see what happens once the training finishes.
  • w

    Waltman13 (semi-active)

    05/22/2021, 11:31 PM
    yeeeeep it's just about turned out the why I expected it. fun..
  • p

    PerroArg

    05/22/2021, 11:57 PM
    retraining
  • p

    pi

    05/23/2021, 12:06 AM
    Because you didn’t read the faq
  • u

    user

    05/23/2021, 12:07 AM
    I did
  • p

    pi

    05/23/2021, 12:08 AM
    How to fix this exact error is in there
  • t

    thebossgamer1232

    05/23/2021, 12:19 AM
    did you put the right txt name? thats all i could think of
  • m

    Manduybandspn

    05/23/2021, 12:30 AM
    I see someone removed Prince Fleaswallow. That was for the better I think, because I think his model was really busted.
  • z

    Zachary(retired f/ voice clonin)

    05/23/2021, 12:37 AM
    Waluigi too
  • n

    NOT ACCEPTING FRIEND REQUESTS

    05/23/2021, 12:37 AM
    Read it slowly. the solution is in there.
  • u

    user

    05/23/2021, 3:23 AM
    anyone know how to remove background sound
  • u

    user

    05/23/2021, 3:25 AM
    wait hold on lemme read the faq
  • j

    Justin

    05/23/2021, 7:02 AM
    Fixed that lil issue: https://github.com/justinjohn0306/Tacotron2-Colab
  • z

    Zachary(retired f/ voice clonin)

    05/23/2021, 12:10 PM
    Working on Filburt, along with Heffer.
  • d

    deathGripsfanforever

    05/23/2021, 12:11 PM
    Good
  • u

    user

    05/23/2021, 2:03 PM
    oh awesome
  • k

    kitsuner

    05/23/2021, 2:54 PM
    So here’s a question that will probably be tough to answer but is there a way to increase the models tolerance for reverb? Because I’ve been making some notes and characters like chef, marge, Tanya, butter cup, bubbles, silver hand; sometimes and others seem to hit these spots fewer the uneven tones of their voice throw the model straight off solely from the fact the character says them in too low or two high a frequency.
  • l

    LindosoLord

    05/23/2021, 3:12 PM
    Training again
  • l

    LindosoLord

    05/23/2021, 3:12 PM
    with the UT3 announcer
  • l

    LindosoLord

    05/23/2021, 4:02 PM
    And wow
  • l

    LindosoLord

    05/23/2021, 4:10 PM
    It sounded awful
  • u

    user

    05/23/2021, 6:08 PM
    (Posted with permission from mega b) I have created a script that will help make creating the
    list.txt
    file less tedious by pre-filling the number of wavs that need to be transcribed. It will output a file such as:
    Copy code
    txt
    wavs/1.wav|
    wavs/2.wav|
    wavs/3.wav|
    wavs/4.wav|
    wavs/5.wav|
    wavs/6.wav|
    wavs/7.wav|
    wavs/8.wav|
    wavs/9.wav|
    wavs/10.wav|
    A prerequisite to using this script is that you have NodeJS installed. You can then run the script in command by running
    node WavScribe
    . Ping me if you have any issues!
  • p

    pi

    05/23/2021, 6:09 PM
    there are already templates
  • p

    pi

    05/23/2021, 6:09 PM
    so no need to install nodejs
  • a

    Awesome Face (GTX 1660 Ti)

    05/23/2021, 6:09 PM
    but do you have even 10k line?
1...115116117...1068Latest