KVR Audio

Yadrichik_Chaya · Post by **Yadrichik_Chaya** » Wed Jun 29, 2022 2:18 pm

The video below has to do with visual art, you just write something and get the text as an image.

”Gothic art of two racoons in purple bikinis eating donuts” and you then get an image of this…

So this means that in a few years AI can scan all the music out there, not only on Spotify and Youtube, but also lesser known artists on Soundcloud.

Make a data profile of their style and then someone anyone can write in text:

”D Locrian Mode, Tablas in 200 BPM, Fretless bass groove in 100 BPM similar to Jaco Pastorius, Viola arpeggios (Pizzicato) with autopan, Chord progressions in the style of gospel keyboardists and finally a fast flute solo with an harmonizer effect played in the same style as video game influenced musicians found on Soundcloud from year 2014 to 2018...”

And then get something like the text above as an audio file…That's pretty scary!

What are your thoughts?

Yadrichik_Chaya · Post by **Yadrichik_Chaya** » Sun Sep 11, 2022 6:25 am

BertKoor · Post by **BertKoor** » Sun Sep 11, 2022 7:43 am

The AI is not coming up with original ideas.
It merely executes yours.

I've played around with dall-e mini that makes images from prompts. You either accept or rejects its outcomes. There's no option for "yeah nice, but change that detail"

chagzuki · Post by **chagzuki** » Sun Sep 11, 2022 9:34 am

What's happening in the visual domain is astounding. I think this is one of those watershed moments, the world has fundamentally changed and as a result everything will be revalued/reevaluated in the light of a new technological norm. Like the invention of the camera etc.. I'm not so sure that these sorts of changes will reconfigure the world of music as quickly... but it's bound to happen to some degree. But in a way it makes sense, or feels like a logical next step in light of what has already happened. We've moved from successive eras where there was still enough scarcity in both the production of music and the limiting factors in the broadcast media in terms of distribution to keep the model of the artist intact, to an era where there's abundance in production of music and methods of distribution such that there's simply more stuff than anyone knows what to do with.
It makes sense that a new form of intelligence can absorb all of this and process it in ways which completely change what's possible.

BertKoor · Post by **BertKoor** » Sun Sep 11, 2022 1:44 pm

In other areas like text, the AI results are interesting. But it lacks soul and purpose. This is due to how it works. Feed it piles of stuff, and it can produce more piles of that stuff.

Put a chimpansee at a typewriter. Given enough time and paper, how long does it take for it to produce the complete works of Shakespear?

Not there yet, but we're getting closer. Abstracy poetry, it can write that. Perhaps scenarios for Peppa Pig or likewise chewing gum for the eyes. Agatha Christie level? Nope. In short texts written by AI there are too many "wtf" moments. Syntactically correct, but the words make no sense.

vurt · Post by **vurt** » Sun Sep 11, 2022 1:48 pm

like finnegans wake.

chagzuki · Post by **chagzuki** » Sun Sep 11, 2022 1:56 pm

BertKoor wrote: Sun Sep 11, 2022 1:44 pm In other areas like text, the AI results are interesting. But it lacks soul and purpose. This is due to how it works. Feed it piles of stuff, and it can produce more piles of that stuff.

In the visual domain there's a creativity in the fusion of an insanely complex amount of learned elements, trained according to what people find pleasing. The humans involved in the training don't have to know why they like a certain result, they just have to tell the AI that they do and it finds deep underlying structures statistically likely to produce a pleasing result.

The massive complexity of what an AI can do in many respects is already miles ahead of what nearly all illustrators are capable of, and the results are near instant.

Yadrichik_Chaya · Post by **Yadrichik_Chaya** » Sun Sep 11, 2022 2:28 pm

BertKoor wrote: Sun Sep 11, 2022 7:43 am The AI is not coming up with original ideas.
It merely executes yours.

I've played around with dall-e mini that makes images from prompts. You either accept or rejects its outcomes. There's no option for "yeah nice, but change that detail"

There is the possibility to change details with Stable Diffusion, check this out:

i need Help · Post by **i need Help** » Sun Sep 18, 2022 6:13 pm

If/when AI is at a level that I feel “threatened” by (as in it can make my genre of music close to, as good as, or better than I can), and assuming I’m still making music of course, I kinda figured I’d just slap a “100% human made” label on my work and then call it a feature lol.

I’m kinda surprised no one in the visual art world has done this yet given the stink AI art is starting to make.

I’m sure someone somewhere will enjoy, support and pay for non-AI music, just like people pay more for something knowing a fellow human hand-made an item vs it being mass produced for less.

vurt · Post by **vurt** » Sun Sep 18, 2022 6:24 pm

what about music written by ai, but played by humans?

Yadrichik_Chaya · Post by **Yadrichik_Chaya** » Mon Sep 19, 2022 9:09 am

Yadrichik_Chaya · Post by **Yadrichik_Chaya** » Mon Sep 19, 2022 9:10 am

24/7 AI improv music, everything from Coltrane jazz to Death Metal:

BertKoor · Post by **BertKoor** » Mon Sep 19, 2022 11:10 am

Can you sue a NeuralNetwork / AI for plagiarism?

Yadrichik_Chaya · Post by **Yadrichik_Chaya** » Mon Sep 19, 2022 12:37 pm

BertKoor wrote: Mon Sep 19, 2022 11:10 am Can you sue a NeuralNetwork / AI for plagiarism?

Good question!

Then again AI robots already bought drugs/weapons etc. like 7-8 years ago and got away with it…
https://www.bbc.com/future/article/2015 ... egal-drugs

BertKoor · Post by **BertKoor** » Thu Sep 29, 2022 9:14 pm

The next step is ready: text to video !!!
https://makeavideo.studio/

AI music created from prompts?