Where Is It On Uberduck? @TheRoyalRuby2000

OMG CAN YOU GIVE ME A MINUTE?!? GOD DAMN!!

There, it's on Uberduck. I'm doing so much stuff at once so that's why it took a while.

Okay

THX Man

@PixPrucer can you send me the model and i'll see what i can do in terms of the pronunciation

I mean the pronounciation is fine, it's just that it appears the synthesis is scuffed

I thought you were having issues with it turning back to latin letters I think i can fix that but i'd need to look at it

I'll send you themodel shortly then

No problem. Now since you contributed, you need the contributor role which I don't have permission to give you sadly.

And How Do I Be A Moderator? (Not Begging)

Have months of experience with being in the Uberduck community.

I Don't Have Money

Who said anything about money?

You don't need money to contribute to the site. LOL

And How Do I Have Month's Of Experience

I think we've got enough moderators for now, but to answer your question, you simply have to create voice models and submit them to Uberduck, and you have to be more involved in this server's conversations.

K

From a friend of mine on Twitter: "I've tried to make another AI voice, but Tacotron 2 is giving me Tensorflow-related error messages again. Do you think you could find a Colab notebook that fixes this issue?"

Use this notebook, it has that issue fixed https://colab.research.google.com/drive/1OFhKJF846EbFsvdNVkapSZvXc7wHvoSd

It worked.

Also from the same friend of mine on Twitter:

Typical checks 1) is the model set to public 2) did it install correctly, which considering it's tensorflow I usually wait a bit and try again which usually works 3) is it even the rigjt nb

Use this notebook ^^ https://colab.research.google.com/github/justinjohn0306/FakeYou-Tacotron2-Notebook/blob/main/Latest_Tacotron2_and_HiFi_GAN.ipynb

Is this pipeline or legacy? FYI this is from the Uberduck guide seen in #842841796568547414, and I've checked the notebooks in app.uberduck.ai and they are different from this specific one

I got this warning saying "warning: audio amplitude out of range, auto clipped." Is it still okay to continue training? (pipeline)

yes

I get that warning and it comes out perfectly fine for me

very late but it probably means they should toggle

pronounciation_dictionary

either on or off (depending which option they used last)

I used the pipline Tacotron and uploaded my zip onto

UPLOAD_HERE

, but when I run Load Dataset, it shows this error

ERRORS:
Unexpected end of archive

ERROR: Data Error : 119.wav
---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-2-7bfed8493c23> in <module>
    125 dirs = [x for x in os.scandir(target_path)]
    126 if len(dirs) == 1:
--> 127   with open(sorted(glob.glob(os.path.join(dirs[0].path, "*.txt")))[0]) as f:
    128     already_multispeaker = (len(f.readline().split("|")) == 3)
    129 

IndexError: list index out of range

it seems like the opened target location is a empty list? Does anyone know what I should do with it?