Hello Folks,

I have this one idea that you could create an AI with Python and thus create countless projects (beats). Does anyone know anything about FL Studio's Python API?
Is it realistic to program something like that? You would need a lib that can interpret samples and categorize them.
Imagine I write and just make a beat like "Drake 8am in Charlotte" and the script sets up a project, renders an mp3 and I can listen to the sketch. If I like it, I can finish the project.

Let me know what do you think?

 

People, as far as I know, make music because they enjoy making music. What is the point of using an AI to make music? Commercial use is okay but I don't really get the point here....

 

This kind of AI project exists already! A Finnish developer-producer (he has PhD in Mathematics, AI and Computer Science) has programmed from scratch a system called AaltoAivoAI, which has analysed Terabytes of audio, music and MIDI, music theory etc, and now it creates music and exportable music projects, virtual instruments, effects etc, plays them, and renders it as audio file if necessary. It can play external instruments and has Ableton Live Link for jamming. You can teach it with your own music to match your own style.

You can see it in here:

AaltoAivo UI is shown at 2:18 onwards.

 

not yet mr talentless

 

Could be possible with PyFLP from github

 

I have beta tested this for several months. It is not final yet, but it is much better than those Google and OpenAI "text-to-audio" solutions. This creates real multitrack projects for each instrument, MIDI, effects and all. Although it can do "text-to-music" and "image-to-music", its biggest potential is in utilizing music theory and working as an "AI music assistant", because it can learn from user material. As far as I understand, the developer is more interested in researching "AI music intelligence" as a concept, than creating some rip-off "AI text-to-something" service.

 

PyFLP seems to be the lib we need.
Believe me this will be the future. Maybe Image-Line is already working on a something like this.

 

If this is the future...
Then damn...
I'm glad I'm old

I'm not getting at all the want of some artifical intelligence to write the music for you, it's like you're both the beta tester for your own demise and welcome the end of musicianship for some odd reason. Somebody please explain why they think not knowing how the create your own art is not a bug but a feature in this scenario down the line? Car mechanics don't mechanic anymore to repair or rebuild engines, they are mere parts changers, jacking into the car's computer and soon will be superfluous when companies transition to robot work that will replace them like in automated automobile factories where the human workforce is now scant in comparison to machines.

Please fill me in.

 

I have not tested AaltoAivo AI, only followed developer and his progress. I understand that it can create its own virtual instruments like NI Kontakt, so it can multisample existing instruments and create totally new instruments that do not exist IRL. Then it uses its learning in music theory to create multi-track compositions according to the mood (and other parameters) you want. I think he developed this project during Covid lockdown because he was bored! I am very interested to test it when I have time, have to check it in GitHub.

 

I remember a cool extension for FL Studio by an advanced user. About keyswitches and stuff. Also made an equivalent but unfinished for Reaper.
So it must have an API. I don't know in which language. As you know there's all kind of wrappers/bindings for python so I guess it might exist one, even if not native.

 

I had contact with Image-Line. They dont work on any similar feature or have any plans. So if someone is into python I would love to start a open source project on github.

 

Unlike images, I believe that AI will never be able to compete with human in music. And that's because Music is a very different kind of art. Very abstract. No one knows the rules for writing good melodies. No patterns to be analyzed and followed, which is what AI is good at. Great musicians often fail, and that proves it. Everyone here knows how hard it is to make 'good' music. Just 7 notes, but it takes hell a lot of efforts to make good music. I like the way all these things are and I hope it will keep the AI clowns away.

 

I hear much uninspired, hack quality lyrics with music that is only as good as the least common denominator and not of any merit. Correspondingly there are ears that like that kind of thing, and although they might like "better music" still will ingest the swill supplied by the aforementioned which amounts to perhaps a billion dollars a year. AI can compete with those who fall into that category. There's a damn good reason that "fix it on the mix" and voice correction software are such merry bedfellows...overall as a percentage of output the quality suffers and AI is beginning to take up the slack and will only become better at doing an equally shitty job for the deaf, dumb and blind folk who like that kind of thing.