AI Computers writing music

AI Computers writing music
0.0 0


I put this in the iPhone recording post but felt it needed its own topic.

@Jonathan mentioned Band in the Box which has an AI melody generator and I thought that he and others would be interested in this new company which is marketing AI generated original music. Interesting stuff!

Here’s one of their pieces.

I Knew Someone would do it One Day

The licensing side of the music seems like it’d be a nightmare. I mean does anyone actually “own” AI created songs??


Hi holster, that’s a good question and I certainly don’t have a good answer.

The Aiva website makes a point of referring to their technology as a person.

" I am The world’s first virtual artist recognized by an author’s rights society"

So they seem to be trying to establish their technology as an entity with ownership rights to be recognized as such by the industry and society as well; which seems to me to be kind of a stretch.

I think in reality this process is just another tool, which means that the person who owns the software and enters the parameters and whatever else is necessary for the program to produce a piece of music should have ownership rights to the work done. It’s not like AI (in this case at least) created itself, then decided what music to write, wrote it and then published it. Someone else is controlling the entire process and should take responsibility for and reap the benefits of the work done.

But who knows, we are on the cutting edge here.


I feel so conflicted. Because, on the one hand, I know I shouldn’t like it, it’s canned, artificial, blah blah blah. …but, I kind of did. :face_with_raised_eyebrow:


I like it too actually. I kept expecting someone like Josh Groban to start singing. Lol


I’m curious to know how much human massaging is going on in these tracks. Are they fully composed and orchestrated by the AI, or is it just spitting out chords and melodies and then orchestrated and programmed by a human?


Solid point


I don’t know the answer to this, but . . . it seems to me that almost any well known and established style of music follows certain patterns pretty closely. So those patterns: melodic, harmonic, rhythmic, formal, arrangement details and orchestration details can be organized into a data base. A computer can easily listen to all of Bach’s work and analyze it in those categories, although some information would need to be entered by humans.

The programmer sets a series of parameters, maybe length, style, key, tempo, instrumental group size and type, dynamic range; all things that the customer will probably have preferences for.

All of this information really gives a pretty narrow pathway for compositional choices to be made, whether by a machine or a human. So then a score is produced. Some human composers are really quick, but we know who will win that race. The machine will work tirelessly 24/7 ; this is a pattern we’ve seen many times before in many fields.

Now to produce the actual music is a separate process but it could easily follow the same pattern. The demo piece on the Aiva website sounds good enough for general use I think, but the stuff you hear on the VI website done by humans is mostly better in terms of sound quality, so I believe the recording/production on the Aiva piece is automated as well.

There would have to be quality control done by humans, and probably certain phrases or sections would have to be redone or maybe even (gasp) done by human hands?

The kicker would be to ask the computer to produce a work that is highly original and incredibly popular, like ‘Sergeant Pepper’. Probably impossible for now at least.


The overall orchestration is pretty nicely done…
The lead melody sounds flaky though…


AI is something that we don’t like because it will likely take work away from human beings, it may be a threat to our independence and safety, it just seems unnatural and soulless.

I think we need to accept it and learn about it, maybe even learn to use it. Kind of like drummers in the 80’s when drum machines started to get popular. Most drummers hated them, but some of them became drummer/drum programmers and continued to get work that way.