Generative.fm

[drkmttr]

I've been a bit obsessed with idea of generative music (or any art) lately. A few months ago I discovered Generative.fm and really enjoy listening to the various pieces. It's especially nice while working, meditating, or falling asleep.

If you're not familiar with generative music, this is a good primer by the same developer of Generative.fm. Basically, the music is generated by your computer, doesn't repeat, and will play as long as you want to listen.

It's still a fairly young project (I'm a patron) and the developer is continually adding new pieces and improvements to the site. My favorite pieces so far are "Awash" and "Day/Dream". What's especially cool about Day/Dream is that the mood of the piece changes depending on the time of day.

With the awesomeness of Peppermint and Ice, you can create an SSB for it. 

[PCNetSpec]

Interesting concept, and not one I'd really come across before .. TVM

[drkmttr]

As a fan of both electronic music and ML/AI, I find this very alluring. It'll be interesting to see how far Alex takes this project. He has mentioned some of the performance barriers with JavaScript so he's considering creating the pieces in WebAssembly.

[stevesveryown]

Wow, very cool!  I'm gathering one can play the music as background in a stream and not worry about copyright?  Thanks for bringing this up!

[drkmttr]

Exactly! That is unless your computer decides to copyright the music. 

Generative systems are a lot like windchimes and your computer is the wind. Although, I think as time goes on I wouldn't be surprised to see true AI created music.

[Slim.Fatz]

Generative systems are a lot like windchimes and your computer is the wind. Although, I think as time goes on I wouldn't be surprised to see true AI created music.

Hi drkmttr,

I take it that you are quite a bit younger than I, so I thought that I might provide you with some background to computer/AI-generated music. I think that you might be surprised to learn that computer/AI-generated music has its roots back in (at least) the 1970s !! 


Back in those "dark ages" before personal computers or smart-phones (this was in the early 1970s), I was a student at the University of Illinois, majoring in mathematics and also very interested in computers, music and music composition. I was very lucky to get to know and become a student in the advanced music composition classes taught by Prof. Herbert Brün, who was teaching (together with Prof. Heinz von Foerster) courses dealing with diverse topics (such as cognition, composition, cybernetics and others) at the University's Biological Computer Laboratory. In order to keep this post short I will refer you to the following links so that you can get an introduction to these two brilliant men: Herbert Brün & Heinz von Foerster. 


These two outstanding people had a profound influence on my development both personally and professionally. Unfortunately, many of their achievements and pioneering work seems to have been forgotten -- most likely due to their intense interest in their research and disinterest in promoting themselves (in contrast to so many people today in the age of lunatic social media self-promotion). I continue to wonder at the claims of younger people about how (supposedly) "new" AI advances and achievements are appearing, when in fact many of these were initially developed and researched back in the 1970s. The extreme difficulty and expense (back in those days) to use or have access to powerful computers was a tremendous handicap. Things have, of course accelerated with today's more powerful, inexpensive and compact computers and computing devices; nevertheless most of these "wonderful and new" AI developments actually have their roots before most of the spell-bound fans were even born. 


Regards,

-- Slim

[drkmttr]

Hey Slim.Fatz!

Thank you for the info and your interesting background. I'll be sure to check out those links you provided to your past professors. You're right, a lot of work was done back in the days to create the mathematical underpinnings of AI and machine learning. Now that we have much more powerful hardware to run it on, it's progressing very quickly. Most things in this world are usually incremental improvements to previous knowledge. However, when I say "true AI created music" I don't mean what we have today or what you had in the 70's. I am referring to true intelligence that can make beautiful music like any human with little to no guidance. I don't mean statistically varied samples and arrangements, I mean the AI understands music theory and applies it with any sounds or instruments provided and the results match or exceed human quality and creativity. They might pull this off with narrow AI but considering the diverse cultural aspects of homo sapiens, it would probably need to be a general/strong AI that would eventually learn music and "feel" it like we do. Since we're still far from such an AI, we don't have to worry about musicians and artists being out of work just yet. 

[drkmttr]

Oh, on a side note, Bandcamp just featured some great artists and generative music. One of the artists is Bing Satellites who I have been a big fan of for quite some time. He's also friends with Cousin Silas who I created a Wikipedia page for.

[Slim.Fatz]

Hey Slim.Fatz!

Thank you for the info and your interesting background. I'll be sure to check out those links you provided to your past professors. You're right, a lot of work was done back in the days to create the mathematical underpinnings of AI and machine learning.

[...]

However, when I say "true AI created music" I don't mean what we have today or what you had in the 70's. I am referring to true intelligence that can make beautiful music like any human with little to no guidance. I don't mean statistically varied samples and arrangements, I mean the AI understands music theory and applies it with any sounds or instruments provided and the results match or exceed human quality and creativity. They might pull this off with narrow AI but considering the diverse cultural aspects of homo sapiens, [...]

Hi drkmttr,

Thank you for your reply. I have added emphasis to several phrases in your reply and will address them in order below.

1. with little to no guidance -> Of course, I cannot expect you to be familiar with the work of Prof. Brün but I wil tell you that is computer/AI-generated music compositions were pretty much exactly that. And that is what was so pioneering about his research. It also triggered quite a harsh backlash from traditionalists. 


 2. music theory -> Here you run into the problem that music theory is not has standardized and the same all over the world as most people believe it to be. What you said yourself, "... considering the diverse cultural aspects of homo sapiens ..." really hits the problem on the nose. 


3. quality and creativity -> Two more very difficult to define and analyze aspects that philosophers, artists, AI researchers, etc. have yet to agree upon. Using these terms sounds great, but actually measuring them is another problem that seems to defy solution.


Regards,

-- Slim