any artists with a music background and a lot of experience prompting Midjourney/Stable Diffusion want early access? 👀
(vqgan+clip or disco diffusion experience is a plus!)
I generated this audio with a text prompt using Grimes Elftech Bot (still in private beta)
The prompt was "synthpop, otherworldly pulsating synth, heavy bass, complex layering"
Honestly, kinda bangs.
Drop some simple prompts in the replies and I'll run them and share results.
@Tantacrul
@Meta
it seems like all meta apps have dark patterns. they're getting really subtle with it too. like no mark you cannot have my phone contacts
AI voice models are the new autotune and your favorite artists will be using them soon
not stylistically/to sound like other artists, but to sound like themselves
their perfectly recorded and edited selves. in one take. with an iphone mic
@Austen
yea. tbh i think using your platform to bring attn to behavior like this is great (not sure posting his name was necessary tho…)
either way i hope he learns from
the experience. if he can put the same energy into his work as he put into the post the kid could def have potential
Making and running ktt as a kid was absolutely ridiculous. Sometimes I regret it… but the positive words I sometimes get from people is so meaningful
My life has been so bizarre so far. And that was such a crazy era — I should probably write a book one day. or see a therapist
@yacineMTB
but yacine. how can dingboard dot com only cost 12 dollars when adobe photoshop costs literally millions of dollars and doesn’t even have any of dingboards powerful features?
The “GPT 4.5” blog post (which reads like a generic placeholder) was last crawled by Bing in September 2023
Unless I’m missing something it seems most likely that an OAI employee accidentally published a draft post (many such cases) and the search result went unnoticed until now
If u r a god and can create a stone u cannot lift, ur failure is not that the task is impossible, but that u created a universe where stones are objects to be lifted. U must introduce more dimensions. I am very fascinated by this thought experiment today.
why does no one use the anatomically correct heart emoji? three medical doctors submitted a fifteen page proposal to the unicode consortium to get this thing. show some respect🫀
Pre-process prompts for text-to-image AI models like Stable Diffusion with large language models (fine-tuned on writing good prompts) as part of a T2I pipeline
I'm testing this in production and the results are usually MUCH better. This is seriously low hanging fruit.
Turns out you can use MusicGen for continuations of *audio*… like you just record a few seconds of an instrument and it'll continue, with infinite variations based on it
BUT if you add text descriptions too, it starts to get insane…
Here's some examples
7 sec original input:
@Grimezsz
@GRIMES_V1
😳
@GRIMES_V1
maybe grimes is okay with it but i dont kill me…
remember we’re v good friends and still have to make all that music and stuff we always said we were gonna make (im usually more productive while alive)