When will we have it anons? The closest thing is kind of Boomy I guess and it sucks balls. I want to get on the music creation AI before it takes off.
Why is there no Stable Diffusion of music yet?
Falling into your wing while paragliding is called 'gift wrapping' and turns you into a dirt torpedo pic.twitter.com/oQFKsVISkI
— Mental Videos (@MentalVids) March 15, 2023
You literally can just use Stable Diffusion for it
https://www.riffusion.com/about
It's not even the state of the art model. There are dozens of music AIs out there. Music was the first art form created by machines even before neural networks. Learn to use google.
I did see this one anon but wasn't too impressed by it.
What would musicians want from such things? Better voice synthesis like Hatsune Miku?
ideally yes, to be able to describe music a little bit more in depth. riffusion seems like it only accepts very general type terms, and has problems getting as detailed as say, stable diffusion would for an image. I don't want to judge it without trying it myself, so I'm running the colab now.
A vocaloid-type thing but with a real voice would be good/neat too. I played around with Vocaloid years ago, but I don't really want a Miku sounding voice. I think most of the tech exists but nobody has made it open source really.
It's shit. Currently nothing is close to SD /Midjourney or OAI in generative stuff.
sex
It will be pretty interesting when generative synths like Phase Plant simply let you PROOMPT instead of doing things the old fashioned modular way.
There will be, and it will suck the life out of the music industry as idiots with no understanding of music churn out meaningless noise 24/7. I'm growing tired of trying to convince retards that current ML applications are nothing more than Dunning Kruger amplifiers that will lead to a draw to entry effect like the world has never seen before. I'm just going to stop paying attention until these retards start crying about the diminishing quality of culture, then I'll return to say, "I fucking told you so. Eat your own shit or starve."
Copyright.
That is what daws and samples already did.
I honestly just want to create songs for myself to listen to really, although if I put them on spotify for free and someone liked them, cool. I'm not that great with music production software. Tried to learn keyboard and guitar but I have smol Trump hands and can't really manage it. Would just be fun to make some shit for my own listening pleasure.
>churn out meaningless noise
That describes rap music as of now in my opinion.
Now imagine all music is of the same quality as rap. Or don't, just wait a few years and see.
All current music is rap already isn't it? I stopped paying attention.
How do you quantify the quality of music?
Whoa, look out! Socrates is here.
Youre just a low iq gatekeeping doomer
The music industry literally cannot get any more lowest common denominator, AI will just break down the barriers to entry and let creative people who are trapped in other fields due to economic reasons (this includes actual musicians) the ability to produce high quality sounds with a higher correlation to what someone can think of with no training in music theory at a velocity that only people with cash could do, AI can give people access to the knowledge of a world class producer / sound engineer or better, what it produces is completely up to the creativity of the person working with said producer.
You act as though language models think for people, wrong, they teach people what is gatekept behind retarded post secondary education systems, money, and networking
honestly i just hum songs to myself i make up a lot but i don't know where to begin really. i tried googling where to even start to just make pop songs at home because that is where i like but there was so much information i didn't even know where to start. no one in my friend group / house is musically inclined or anything, and i didn't know where to even start with music theory and production software. i have little ideas in notepads on my phone, but i never really did anything with them
That's why these shills are going to such lengths to turn people away from AI. Existing gatekeepers can't gatekeep any more.
>GATEKEEPER
What the fuck? Study economics, smoothbrains. I'm not talking about gatekeeping, I'm talking about an economic phenomenon where a lucrative industry is destroyed by too many vultures trying to cash in. There should be barriers to entry based on ability, not previous financial success or social/political clout. When all barriers to entry into a lucrative industry are removed, the financial incentive draws every talentless hack out of the woodwork and produces a flood of generic garbage in the hopes of securing a little slice of the pie. The result is the consumer base becomes stretched too thin to fund any single entity to the point where the entity can afford significant reinvestment into the industry. Innovation grinds to a halt, the industry fails to attract sustained consumer attention, investment is directed elsewhere, the industry collapses, and the vultures (mostly Indians with a sprinkle of other third worlders historically) move on to the next meal ticket. I'm not advocating gatekeeping, I'm only describing a well known phenomenon commonly called the draw to entry effect.
You quoted half a sentence and then misused the word "ironic" so I have no clue what the fuck you're trying to get at. You seem to have some misconceptions about The Dunning Kruger effect, which isn't some kind of mental disorder like you seem think it is. It's something everybody experiences as they become more knowledgeable of any particular subject. The problem with something like AI is that it enables people who are new to a subject to produce results beyond the limits of their understanding, known also as their horizon of doubt. As somebody's understanding of a subject grows, their horizon of doubt expands and they become less confident in their knowledge on the subject. AI assistance short circuits this cycle of learning, doubt -> understanding -> doubt -> etc, resulting in stunted knowledge and severe overconfidence.
music isn't going to collapse because people are listening to things you don't like
>the draw to entry effect
returns nothing for me, you sure you didn't make it up ? peak dunning-Kruger if that were the case.
why do underage idiots who advocate for gatekeeping use a website with no barrier of entry ?
>I'm growing tired of trying to convince retards that current ML applications are nothing more than Dunning Kruger
ironic
i don't care about this, i just want to beep boop bop and make some carly rae jepsen sounding shit for myself to listen to. metalfags are so weird and gatekeepy about music
ew her eyeball is melting and her neck is off center. wtf am i looking at another victim of tyrone?
sorry can't hear you anon
yeah i gave it a try in colab and it sucks for now
Also, as a fan I will NEVER replace our Nana-sama with your shitty neural-network spawned piece of trash Super Sonico copycat, so get back to your slum.
i don't know what the fuck you're talking about weebro, i just made a picture of a girl with pink hair wearing headphones and want to make some of my own tunes to listen to while i vidya.
good thing i am not really looking to compete, i just want to know when the ship is going to set sail so i can hop on it and enjoy the ride and create some fun stuff
nice. top left looks almost real. prompt?
>looks like domestic battery victim
>headphones say 'jamal' on them
stable diffusion is for mudshark loving cucks confirmed. can't make this shit up
I'm not an expert but isn't there are pretty clear rules for music like time signature and other stuff? Music theory describes pretty well how to make things to sound "good". If you make a some sort of model around the music theory you can produes good note sample. I think it is possible to do that just by using ChatGPT. Then just use classical instrument samples for generating the sound using those notes. Creating a good electronic music would be harder challenge probably.
so in my head, you could maybe have the AI trained on specific genres, but also specify bpm, outline where you'd like certain things to take place in the song. maybe specify at what timestamp the chorus or bridge should be, etc. basically kind of paint it with words and describe it mostly, and then it brings it to life.
i'm looking for like, creating loras and such i guess, but not of one musician/artist's work. like the difference between using lots of loras and inspiration for a SD image rather than just "scary girl in the style of junji ito".
so like i'd want to be able to specify i want a song to be this long, sound like hyper pop, specify bpm etc and have the AI create it. you're right that music already has a lot of "rules" for sounding good, so i think someone talented enough could just make something follow said rules.
"In this world, even a Soviet one, there will always be rich and poor. Rich in gifts, poor in gifts. Rich in love, poor in love."
Can you give a live concert without playback? No? Fuck off!
We already have it but no AI-fag can compete with real-life cute idols and batshit crazy musicians:
Come on dawg, try to compete with these, you can't? What is that? Your little shitty program still needs 2 more weeks to compensate for your lack of talent?
Because the music industry has protections against abuse, to a fault
Would it be possible for someone to create one that is designed solely on royalty free music? There's tons of royalty free music for streamers it could be trained on noe
You can already do chords. Making full music sheets for entire orchestra will be also a piece of cake. However generating an actual sounds and music that does not sound like a children's synthesizer will be a huge challenge, it's harder than TTS.
>Why is there no Stable Diffusion of music yet?
there is & has been for a while now - https://openai.com/research/jukebox
precedes people 'making' pics to wank to akshully.