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

    game time 2 (forgottenPW)

    08/05/2022, 2:15 PM
    the synthesis notebook doesnt work anymore
  • g

    game time 2 (forgottenPW)

    08/05/2022, 2:16 PM
    halp
  • j

    jonny5alive

    08/05/2022, 5:04 PM
    Try the new ones here I still get 'ValueError: Expected more than 1 value per channel when training, got input size torch.Size([1, 512, 1])' on the Tacotron Pipeline though
  • j

    jonny5alive

    08/05/2022, 5:24 PM
    Ok I fixed it I am best
  • j

    jonny5alive

    08/05/2022, 5:24 PM
    I have 97 models to train
  • j

    jonny5alive

    08/05/2022, 6:07 PM
    Does anyone know if there are characters that need to be escaped in list.txt? And can multiple files have the same transcript?
  • u

    {K EY1} (Kei)

    08/05/2022, 8:01 PM
    @jonny5alive
  • j

    jonny5alive

    08/05/2022, 8:59 PM
    @{K EY1} (Kei)
  • u

    {K EY1} (Kei)

    08/05/2022, 9:02 PM
    .
  • u

    {K EY1} (Kei)

    08/05/2022, 9:02 PM
    bro i literally provided the answer to your issue
  • u

    {K EY1} (Kei)

    08/05/2022, 9:02 PM
    this issue
  • j

    jonny5alive

    08/06/2022, 12:19 PM
    nah
  • h

    haru0l

    08/06/2022, 12:43 PM
    no that was the fix for the earlier problem
  • m

    MapleNeko

    08/07/2022, 7:00 AM
    i dont know why i feel tacotron pipeline is better than legacy
  • p

    PopcornBubblegum

    08/07/2022, 7:27 AM
    probably because the pipline one is newer than the legacy one
  • m

    matty1406

    08/07/2022, 10:29 AM
    Pipeline is better if you know what you’re doing
  • u

    {K EY1} (Kei)

    08/07/2022, 4:13 PM
    Pipeline also has arpabet but the legacy notebook we have does not
  • t

    TheWu-Cepticon

    08/07/2022, 4:59 PM
    I'm still getting this error
  • h

    haru0l

    08/07/2022, 5:53 PM
    @TheWu-Cepticon
  • h

    haru0l

    08/07/2022, 5:53 PM
    kei literally tagged u the fix before
  • f

    Fargo

    08/07/2022, 6:48 PM
    in training section
  • h

    haru0l

    08/08/2022, 1:39 AM
    use the notebooks in #841437191073955920
  • h

    haru0l

    08/08/2022, 1:40 AM
    that version is super old
  • f

    Fargo

    08/09/2022, 1:21 AM
    Any tutorial for it?
  • h

    haru0l

    08/09/2022, 1:22 AM
    process is roughly the same
  • l

    Linkle Ze

    08/10/2022, 2:02 AM
    I was wondering where to find the two ids in a model for inferencing in the Hi-Fi Gan Synth notebook?
  • g

    Gosmokeless28

    08/10/2022, 5:14 AM
    In the middle of the URL of a Google Drive link
  • s

    Sol ROO

    08/10/2022, 8:52 AM
    Hello everyone! do you know where I can find a sample dataset (wavs + transcription) that gives good results? I would like to have a way of knowing if I am doing the process correctly.
  • a

    a nickname?

    08/11/2022, 5:22 AM
    https://colab.research.google.com/drive/1CIPTTj94EocZe2w5-zCDC3G44K9OaFal#scrollTo=iQBUZTuMogui this notebook is broken
  • h

    haru0l

    08/11/2022, 5:26 AM
    what do you mean broken?
1...262728...158Latest