As a child in the 2010s: I occasionally used Cleverbot and This Person Does Not Exist as fun toys during recess.
Early 2020s: I goofed around with Character AI to see fictional characters “come to life” and to see if AI worked for language learning. I quickly grew bored of Character AI after maybe a day or two, and decided it didn’t really have any practical uses for me. I also used 15.ai for a while — a text-to-speech website trained on various fictional characters — back when that was a thing, and although I had some fun with it and thought it had some nice features and potential, I still felt it was too limited for me to have much use for it, and so I eventually stopped using it, and the website eventually went inactive.
Mid 2020s: I have by this point started using Retrieval-based Voice Conversion and This Person Does Not Exist in personal projects like worldbuilding and fandubbing. I have no other uses for “AI” technologies, certainly no uses for text generation aside from I guess occasional machine translation.
Oh I should have figured it’d be a bit outdated, and it’s not the most safe either since the standard is safetensors files. Stablediffusion is probably the better choice for this type of thing.
I recall seeing you commenting on language-related stuff a lot here, so if you could elaborate on how you tried to integrate it into language learning and why it didn’t work out for you, I’d be happy to read it!
There isn’t much to it, I just tried to use it as a conversation partner and the text it spat out was completely broken. Like it was blending English and Japanese vocabulary and grammar in a really ridiculous way.
Maybe if I tried again now and used a different service, it would work better, but I just don’t think it would be worth it.
My experience with so-called “AI” thus far:
As a child in the 2010s: I occasionally used Cleverbot and This Person Does Not Exist as fun toys during recess.
Early 2020s: I goofed around with Character AI to see fictional characters “come to life” and to see if AI worked for language learning. I quickly grew bored of Character AI after maybe a day or two, and decided it didn’t really have any practical uses for me. I also used 15.ai for a while — a text-to-speech website trained on various fictional characters — back when that was a thing, and although I had some fun with it and thought it had some nice features and potential, I still felt it was too limited for me to have much use for it, and so I eventually stopped using it, and the website eventually went inactive.
Mid 2020s: I have by this point started using Retrieval-based Voice Conversion and This Person Does Not Exist in personal projects like worldbuilding and fandubbing. I have no other uses for “AI” technologies, certainly no uses for text generation aside from I guess occasional machine translation.
Does this have a local version like RVC does?
I’ve been meaning to look into that, but it isn’t something I use super often anyways.
Yes it does, it was an Nvidia trained model. I forget the exact model name though
Edit: https://github.com/NVlabs/stylegan?tab=readme-ov-file
This is what the person does not exist ran, not 100% sure how to run or anything though.
Oh I should have figured it’d be a bit outdated, and it’s not the most safe either since the standard is safetensors files. Stablediffusion is probably the better choice for this type of thing.
I recall seeing you commenting on language-related stuff a lot here, so if you could elaborate on how you tried to integrate it into language learning and why it didn’t work out for you, I’d be happy to read it!
There isn’t much to it, I just tried to use it as a conversation partner and the text it spat out was completely broken. Like it was blending English and Japanese vocabulary and grammar in a really ridiculous way.
Maybe if I tried again now and used a different service, it would work better, but I just don’t think it would be worth it.