been fukking around with the Suno AI music maker

↓R↑LYB

I trained Sheng Long and Shonuff
Joined
May 2, 2012
Messages
44,204
Reputation
13,743
Daps
171,150
Reppin
Pawgistan
we gon see, we've had virtual bands before of course
This is beyond a virtual band. The generative abilities of these models is something the world hasn't seen before. Suno.ai hasn't even been around for 3 years.
New music will be anti rhythm so the ai cant figure it out

The real artists will thrive. The fakes will fall. Cant wait
These models can figure it out. That's literally what they're designed to do.They're universal function approximators. If they can figure out how to simulate global weather patterns they can figure out Silk the Shockers rhyme pattern.

The Universal Approximation Theorem​

Mathematically speaking, any neural network architecture aims at finding any mathematical function y= f(x) that can map attributes(x) to output(y). The accuracy of this function i.e. mapping differs depending on the distribution of the dataset and the architecture of the network employed. The function f(x) can be arbitrarily complex. The Universal Approximation Theorem tells us that Neural Networks has a kind of universality i.e. no matter what f(x) is, there is a network that can approximately approach the result and do the job! This result holds for any number of inputs and outputs.

1*MxlOQO3vXtD1Q1swLP-Wqw.png
 

Jesus H. Christ

I died for your sins
Joined
May 3, 2012
Messages
16,502
Reputation
3,846
Daps
59,700
Here's a Mexican corrido intro dedicated to the honorable Tariq Nasheed. It might be over for brehitto musicians.

images


 

Bugzbunny129

All Star
Joined
Apr 14, 2020
Messages
2,636
Reputation
957
Daps
6,066

:scust:
A little better
Literally gave me the same lyrics. Hmmmmmm. Lol
 

Bugzbunny129

All Star
Joined
Apr 14, 2020
Messages
2,636
Reputation
957
Daps
6,066
This is beyond a virtual band. The generative abilities of these models is something the world hasn't seen before. Suno.ai hasn't even been around for 3 years.

These models can figure it out. That's literally what they're designed to do.They're universal function approximators. If they can figure out how to simulate global weather patterns they can figure out Silk the Shockers rhyme pattern.
You can fugure it out but you wont know to do that first. It needs a base first to work off of. It wont know how to make certain type of songs until theyre made.

I doubt these things can make real sounding boomvap for a while with real sounding dusty “samples” Other shyt sure. When it can itll be super simple. Its gon be sometime before it adapts correctly.
 

↓R↑LYB

I trained Sheng Long and Shonuff
Joined
May 2, 2012
Messages
44,204
Reputation
13,743
Daps
171,150
Reppin
Pawgistan
You can fugure it out but you wont know to do that first. It needs a base first to work off of. It wont know how to make certain type of songs until theyre made.
That's not true. These models don't just train on human generated data anymore. They also train on AI generated data as well.

Reinforcement Learning from AI Feedback (RLAIF) is a method devised by Anthropic that is designed to overcome many of the shortcomings of RLHF. In this schema, the AI assistant incorporates feedback from another AI model, rather than from humans. Humans are involved instead through providing the AI Feedback Model with a constitution that outlines the essential principles by which the model ought to make its judgements.

I doubt these things can make real sounding boomvap for a while with real sounding dusty “samples” Other shyt sure. When it can itll be super simple. Its gon be sometime before it adapts correctly.
What are these assumptions based on?
 

Bugzbunny129

All Star
Joined
Apr 14, 2020
Messages
2,636
Reputation
957
Daps
6,066
That's not true. These models don't just train on human generated data anymore. They also train on AI generated data as well.




What are these assumptions based on?
Every single one ive seen not being able to do what they say they do… the one above works in the trap sound. That “wu tang” beat sounds NOTHING like a real rap beat at all.

Ive yet to hear a dope boombap soundinh “sample” beat from these things that didnt sound fake

I didnt say it wont happen. It just will take time. And when people stop making same song over and over itll be tougher to create “new” things.


Its called “AI” but its not actual intelligence. It can only make what it can study. It can throw diff things in new forms but it still needs those forms to do stuff.

So if kanye makes college dropout it can turn that into anything

But only kanye could make yeezus. Ai wouodnt be able to create that from college dropout.

One day yes 100% but we aint nowhere close to THAT

Butit will 100% be a tool artists use no matter how cheap it is.
 
Last edited:
Top