#Looking for Piper GPU training donation: GLaDOS voice

1 messages · Page 1 of 1 (latest)

spark pebble
#

I tried to train a Piper voice based on the game assets from Portal2. It turned out okay-ish, though my GTX1660Super really limits the batch size and training speed. I sadly don't have more compute power available nor the time to leave it running in the background.

Here are some examples: [https://go.wetransfer.com/t-nhqOLcDQkh]

Her voice makes hiding imperfections pretty easy. Since we can't acquire more recordings, the only option left, is to change the training process.
If someone with a better GPU want's to give this a shot, I can send over the dataset, which I have already prepared. Since I don't know about the legality of sharing these fiels (dataset and onnx) I will not be posting them publicly, as I don't own the copyright (though they are available on the internet).

restive void
#

I remember seeing someone use audio filters and whatnot (think autotune) to get a very glados sounding voice. I wonder if something like that might be an option to generate more samples. Or to just create a processing layer that uses a different default voice and adds a filter on top.

#

I would absolutely love a glados voice.

spark pebble
spark pebble
restive void
#

Yeah, but if you could get a bunch of raw audio samples from her in the glados style, then the processing could be added on. She might even have some idea what processing was done.

#

Might be the most expensive option though

#

Also not sure how a voice actor would feel about thier performance being immortalized in AI.

#

I put Light turned on. and I see someone put Lights have been turned on.

spark pebble
restive void
#

I think the quality seems pretty good, but I guess it's been a while since I've heard the original

#

Ahh I see what you mean

stiff lodge
#

@spark pebble I'm assembling an STT server based on GTX1070 8GB. Not sure how HW compares to what you have, but I can leave it running for a while if that will help. LMK. Would love to be able to use GLaDOS voice for STT

deft epoch
#

Hi @spark pebble i am also trying to make a model for GLaDOS but i'm quite new to linux and stuff like that. I use WSL on windows 10 and i am getting errors with the preprocess. I don't know if it's because of my dataset or if i'm making something wrong with the commands. My gpu is a 3060 12gb.

spark pebble
deft epoch
#

I had already seen those but from what I have seen it's from the dataset I found that's not working. Is it possible to get your dataset?

deft epoch