Google created an AI-based, open source music synthesizer
16.03.2018 Move over musicians, AI is here. Google's 'NSynth' neural network is designed to take existing sounds and combine them using a complex, machine learning algorithm. The result? Thousands of new musical sounds, and an instrument you can play them on.

SYNTH AI - OVERVIEW
2.395 Aufrufe 24.08.2022 #synths #musicproduction #virtualinstruments

Learn more: www.samplelogic.com/products/synth-ai/

 

I just tried this and the pictures are hella blurry lol

Try listening to MF DOOM's mm food

 

Stairway to Heaven - Led Zeppelin - But the lyrics are Ai generated images

 

I got goose bumps. It's like a mix of Pink Floyd and Led Zeppelin.

 

Images are the first step, wait until we have animation... Or VR animation, it will be like dreaming.

 

Absolutely gorgeous!
How long would an human artist need to do a similar work and how much would it cost?
Would it be another field for AIs (creation of animated video clips through images conceived especially for the product)?

 

Last edited: Sep 11, 2022

There's clearly still a great deal of human input required so I imagine it would be something an artist could charge a lot of money for right now, probably for some considerable time. I used to do some work in 3d and it's highly specialised, very time consuming, I envision AI will be a great boon to 3d work.

 

You made me think of something. And that is that the AI will always need the human being to "Update" because the human being will always imagine beyond things. Human beings have a limitless appetite for wanting more and more. That is, the AI may learn to make videos by itself, but then the human being will want something else, and something else, so it will probably always be upgrading the AI

 

The nature of AI consists in consolidation of all trivial and non-creative practices. The most common practices. This is as I've said the nature of AI. Feed more data and get the most common patterns.

AI is good for achieving 'standard' tasks.

The single creative application of AI is when it doesn't work properly. One good example (the video has been made by AI):

 

Time Pink Floyd but an AI attempts to continue the song (OpenAI Jukebox)

 

[A.I.] - "Brick in The Wall" - Pink Floyd - Artificial Intelligence attempts to continue the song

 

Throughout most of human history... Humans have developed technology that has simplified formerly tough tasks. There has long been a strong contention that computer music is not real. Electric guitars, on the other hand, are not true music. With more personnel like wealthyliving.com in the area and a simplified overall procedure. True artists find inventive methods to push that area further than it has ever been pushed before. This is fundamentally the artist's/job. Creative It is to push the provided set of current-day tools to their boundaries.

 

Last edited: Oct 28, 2022

Can Ai help by adding a bit of humanization to my 4 on the floor beats?

 

Yes, for your very soul and your spawn's total future prosperity in return, of course.

 

Not sure if you've actually been listening to mainstream music in the last few decades, but I can assure you music production has totally upped it's usage of autotune, badly abused autotune, glitching, re-pitching and a whole array of weird and wonderful fx, to alter your perception of what a human voice is.

In the same way that it's easier to globally redefine what music is, thus massively lowering the bar of entry & quality, than it is to design an AI to rival Mozart or The Beatles ......... it's much easier to globally re-define what passes for the human voice than it is to develop, in the short-term, a 100% accurate facsimile.

 

Google's getting pretty close now.

https://google-research.github.io/seanet/musiclm/examples/

 

I love Underbelly's (humorous) take on the topic: