📣 **announcement** 📣 ever since i joined uberduck i've been disappointed by its utterly glacial rate of progress after over a year of still having to suffer through tacotron and talknet i've decided to take matters into my own hands i have been abusing both kaggle and colab 24/7 for the past several months to create *the ultimate model* - tens of thousands of realistic voices from everything on netflix and more - exclusive high-quality impressions for blacklisted voices - billions of entirely original synthetic voices waiting to be generated - finetunable on 5 seconds of data, more accurate than elevenlabs - perfect intonation thanks to the dedicated Clustered Bidirectional Tone engine (or use your own reference audio) - downloadable, and packed to a mere 420mb using middle-out compression - runnable in real-time even on mid-range phones - free, forever the model is still a few weeks away from completion, but as a parting gift to the uberduck community i'm sharing a preliminary version early enjoy http://beta1.octopusbot.net/

Thx for the promo

This video is what first introduced me to AI: https://www.youtube.com/watch?v=KSdU5B_fdVo

how do i get torchmoji

you enabled gpu in the notebook but not in the runtime settings

runtime → runtime settings → change from none to gpu

uhhhhhh

did you run the first cell

after the change

https://colab.research.google.com/github/Mildemelwe/Non-English-Tacotron-2-Training-Notebook/blob/main/Non_English_Tacotron_2_Training.ipynb#scrollTo=rJwhEnN9Ddsb

Has anyone been able to run this notebook at all or do you get the Tensorflow error too?

For being Tensorflow 2.0 and not 1.15

I'll dm you soon

Made it compatible with tensorflow 2.0

Here's Pipeline Optimus at 160 epochs.

Back to getting errors that are entirely on me

I'm getting the old "audio file not found" error despite it being uploaded so something is wrong with my workspace

Let's see, I converted the transcript using the proper Japanese notebook and used that setting while uploading the transcript

Penny (Wario Smooth Moves); VA: Ann Lin 300 epochs ~

Hi!I offer some job opportunities to use my voice and my face to make some music videos and I will pay for it how to do it please. thank you so much!

Prime at 180 epochs. I actually trained him to 200 and will test that iteration later. I stopped training there because his loss was starting to go up again.

<@161352382918164480> Too many attachments

I was about to say "Mute it, Lizard-Brain!" to Dyno-Bot just then. XD

Ashley (Smooth Moves) 460th epoch

Ashley (Smooth Moves) 1200th epoch ID: ashley-smooth-moves

https://www.iresizer.com/

i don't think this is machine learning

the way photoshop did it doesn't use any neurals

and it says 2011-2020 so it's an old