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

    user

    05/03/2021, 8:32 PM
    Not quality, just the bare minimum
  • u

    user

    05/03/2021, 8:33 PM
    I hope that's okay to ask
  • s

    SidPlays_144p

    05/03/2021, 8:33 PM
    I don't know the exact bare minimum, but Phil Swift I believe only has 8 minutes of data used for his model on the site and it sounds pretty good
  • u

    user

    05/03/2021, 8:35 PM
    Ooh!
  • u

    user

    05/03/2021, 8:35 PM
    Interesting!
  • s

    SidPlays_144p

    05/03/2021, 8:35 PM
    You'll generally get better models with more data, though
  • p

    pi

    05/03/2021, 8:38 PM
    10 minutes or more for a somewhat ok model
  • p

    pi

    05/03/2021, 8:39 PM
    more than 30 minutes for a good model
  • p

    pi

    05/03/2021, 8:39 PM
    the bare minimum in probably around 3 minutes
  • u

    user

    05/03/2021, 8:40 PM
    Hmm... make that 2:30...
  • p

    pi

    05/03/2021, 8:40 PM
    depends on the voice and how well tacotron learns the alignment
  • u

    user

    05/03/2021, 8:41 PM
    From what I've seen... seems like it responds better to soprano voices than lower ones.
  • u

    user

    05/03/2021, 11:52 PM

    https://i.imgur.com/jUQE9S3.png▾

    AAAAaaAaAAAAAAaaaaaAaAAAAaaaaAAAAA
  • u

    user

    05/03/2021, 11:55 PM
    Just after posting this my PC crashed and I didn't even bother saving before it happened
  • u

    user

    05/03/2021, 11:55 PM
    I hate it here
  • u

    user

    05/04/2021, 12:12 AM

    https://i.imgur.com/Vp3sXPY.png▾

    NOTSOBOT'S OCR COMMAND IS HERE TO SAVE THE DAY THANKS TO THE SCREENSHOTS
  • u

    user

    05/04/2021, 12:31 AM

    https://i.imgur.com/4f7EV01.png▾

    ONLY THREE MORE TRANSCRIPTS LEFT TO FORMAT PROPERLY LESS GOOOOOOOO
  • u

    user

    05/04/2021, 12:56 AM

    https://i.imgur.com/R5jbTZO.png▾

  • u

    user

    05/04/2021, 12:56 AM
    Anyone here who knows Tacotron 2, do you know how to fix this?
  • p

    pi

    05/04/2021, 12:58 AM
    if you want help, you gotta show the error message
  • u

    user

    05/04/2021, 12:58 AM
    I have no idea why it's doing this. All my files are 22.5kHz Mono WAV files and I formatted the transcripts properly
  • u

    user

    05/04/2021, 12:58 AM
    This screenshot has the error message
  • u

    user

    05/04/2021, 12:59 AM
    oh wait
  • u

    user

    05/04/2021, 12:59 AM
    it was cut off
  • u

    user

    05/04/2021, 12:59 AM

    https://i.imgur.com/xSd11Qt.png▾

  • p

    pi

    05/04/2021, 12:59 AM
    only parts of it
  • u

    user

    05/04/2021, 12:59 AM
    There
  • p

    pi

    05/04/2021, 1:01 AM
    as the error message says: it can't find the .npy files it generated
  • u

    user

    05/04/2021, 1:01 AM
    Wtf???
  • p

    pi

    05/04/2021, 1:01 AM
    watch the tutorial again carefully
1...363738...1068Latest