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

    matty1406

    09/02/2022, 7:07 PM
    it started at epoch 1 and 0 steps when i ran it
  • c

    Cris140

    09/02/2022, 7:09 PM
    oh, that's the problem then, this colab already starts at a high step because it uses Crust if I remember well, it probably didn't downloaded the model correctly
  • m

    matty1406

    09/02/2022, 7:10 PM
    but i'm making a hifi gan for legacy not pipeline or does it not matter?
  • c

    Cris140

    09/02/2022, 7:18 PM
    The pipeline code doesn't use the universal model, it uses Crust, so it will start at already high step.
  • g

    Gamma Prime

    09/02/2022, 11:31 PM
    I'm gonna try to ask ONE MORE TIME. IS THIS VOICE POSSIBLE?
  • u

    Uberduck

    09/02/2022, 11:38 PM
    @Krab quacked: hey guys whats up.
  • u

    Uberduck

    09/02/2022, 11:38 PM
    @Krab quacked: hey guys whats up how are you ding.
  • g

    Gamma Prime

    09/02/2022, 11:46 PM
    OK...Why are you quacking in here?
  • s

    samkoelle

    09/03/2022, 12:20 AM
    ahh we kinda nuked the distributed code in the latest updates cause we didnt think anyone was using it - if you pin a previous version it should work
  • b

    Bunitodah

    09/03/2022, 7:01 PM
    so I have a phone, can I do anything with it? like the stuff you people are doing?
  • u

    {K EY1} (Kei)

    09/03/2022, 11:27 PM
    You can, it'll just be more difficult
  • o

    OccultMC

    09/04/2022, 1:45 AM
    https://cdn.discordapp.com/attachments/991602573217890344/1015798654390509659/unknown.png?size=4096
  • o

    OccultMC

    09/04/2022, 3:00 AM
    why isnt the tacotron crust inference notebook working?
  • o

    old inuyasha ꄐ

    09/04/2022, 7:19 AM
    im having trouble on step 3 and 4
  • o

    old inuyasha ꄐ

    09/04/2022, 7:21 AM
    whats it mean by "Import the transcript .txt file into the /content/tacotron2/filelists folder on the left. Training_file: Insert text here"
  • p

    PixPrucer

    09/04/2022, 12:20 PM
    Managed to fix this issue by switching to the pipeline Eng notebook (which had the option for Polish apparently...)
  • p

    PixPrucer

    09/04/2022, 12:20 PM
    The Polish notebook is prety much outdated at this point
  • p

    PixPrucer

    09/04/2022, 2:57 PM
    aaaand now I can't even use my freshly trained model (am using the pipeline inferencing notebook)
  • p

    PixPrucer

    09/04/2022, 2:58 PM
    Hope the model isn't corrupt in any way
  • c

    Cris140

    09/04/2022, 3:26 PM
    You need to set the right symbol set, the same as you trained your model
  • p

    PixPrucer

    09/04/2022, 3:31 PM
    I see There could be an option in the cell for changing that (language choice), but I don't think there's anyone besides me doing anything other than English, so I'll take the effort and go through manual labor
  • c

    Cris140

    09/04/2022, 3:32 PM
    But there is, it's the symbol_set option
  • c

    Cris140

    09/04/2022, 3:32 PM
    I train English, Portuguese and Spanish models
  • p

    PixPrucer

    09/04/2022, 3:33 PM
    I must be using the outdated synthesis notebook then
  • h

    haru0l

    09/04/2022, 3:35 PM
    the symbol_set option seems to be gone?
  • p

    PixPrucer

    09/04/2022, 3:35 PM
    I can't see it either
  • c

    Cris140

    09/04/2022, 3:35 PM
    yeah the option is gone, I didn't knew that
  • p

    PixPrucer

    09/04/2022, 5:37 PM
    now my question is, how and where do I change the symbol list...
  • p

    PixPrucer

    09/04/2022, 5:37 PM
    I can't find the resources to this task anywhere
  • c

    Cris140

    09/04/2022, 5:40 PM
    https://colab.research.google.com/drive/1GM4j91qsq_awvclVC4I2H3fBsWCpcwEF
1...394041...158Latest