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

    Akane

    10/20/2022, 2:46 AM
    Oh, after I reuploaded my .txt file, it is no problem now...
  • a

    Akane

    10/20/2022, 2:50 AM
    Maybe the txt file has 2 blank line at the bottom
  • c

    Cris140

    10/20/2022, 3:06 AM
    Try to remove the :
  • c

    Cris140

    10/20/2022, 3:06 AM
    It may be this too
  • g

    Gosmokeless28

    10/20/2022, 3:23 AM
    But colons are allowed
  • c

    Cris140

    10/20/2022, 3:24 AM
    Sorry, at first I thought that it was talknet training, I don't know how talknet handle these symbols
  • l

    legorunnerkid

    10/20/2022, 2:15 PM
    So now I am getting this "list index out of range" error.
  • l

    legorunnerkid

    10/20/2022, 2:20 PM
    Forgot to mention, this is the legacy version.
  • l

    legorunnerkid

    10/20/2022, 2:26 PM
    Okay, the issue was that I had empty lines in my .txt file.
  • l

    legorunnerkid

    10/20/2022, 2:26 PM
    You just get rid of of those.
  • l

    legorunnerkid

    10/20/2022, 3:11 PM
    So I am having issues now with my model. Now to be fair, this model is brand new and it's not even finished yet (it as about 118 voicelines). I tried to use the Uberduck Tacotron2 synthesizing via pipeline to hear the voice. When I tried to listen to it though, I get Twilight Sparkle's(?) voice. I assume this has something to do with the speaker_weighting.
  • l

    legorunnerkid

    10/20/2022, 3:11 PM
    I've use legacy to make the model btw
  • l

    legorunnerkid

    10/20/2022, 3:18 PM
    I tired the actually_id_like_action_53_instead_please approach and the voice is completely unrecognizable.
  • l

    legorunnerkid

    10/20/2022, 3:18 PM
    Oh wait, I might have been using the wrong one.
  • l

    legorunnerkid

    10/20/2022, 3:33 PM
    So, I'm not sure where I should put this, but I want to show off a result now.
  • l

    legorunnerkid

    10/20/2022, 3:33 PM
    "It's about time I can finally talk."
  • u

    🏝✍Berkeley•Ray•Brewski

    10/20/2022, 6:52 PM
    I used Wifi instead of a Moble hot spot I Used the Chrome instead of Opera GX I even used the Legacy Tacotron Model by Gosmokeless28 ... where am I going wrong?
  • u

    🏝✍Berkeley•Ray•Brewski

    10/20/2022, 6:53 PM
    was it this?
  • g

    Gosmokeless28

    10/20/2022, 7:52 PM
    Batch size 1000?! 😮
  • l

    Lexi (delulu posts on the daily)

    10/20/2022, 7:55 PM
    1000?????
  • z

    Zachary(retired f/ voice clonin)

    10/20/2022, 7:55 PM
    1000 is too high aaaa
  • l

    Lexi (delulu posts on the daily)

    10/20/2022, 7:55 PM
    GIRL WTF
  • l

    Lexi (delulu posts on the daily)

    10/20/2022, 7:55 PM
    WHY 1000
  • u

    🏝✍Berkeley•Ray•Brewski

    10/20/2022, 7:57 PM
    😳💦 ✋🤚 Don't shoot The recommended amount was 1000 with other Tacotron 2 models, I didn't know what was to high so I went with that
  • g

    Gosmokeless28

    10/20/2022, 7:59 PM
    Bro, read the paragraph about batch size I've written there
  • u

    🏝✍Berkeley•Ray•Brewski

    10/20/2022, 7:59 PM
    😨👍👍💦 sure thing
  • z

    Zachary(retired f/ voice clonin)

    10/20/2022, 8:01 PM
    I don't set batch size to 1000 when I train voice models
  • u

    🏝✍Berkeley•Ray•Brewski

    10/20/2022, 8:21 PM
    What was the size when you did your batch? 📃✍️😳💦
  • z

    Zachary(retired f/ voice clonin)

    10/20/2022, 8:23 PM
    11-21
  • z

    Zachary(retired f/ voice clonin)

    10/20/2022, 8:23 PM
    each voice models I made have different batch sizes
1...565758...158Latest