music21 VS jukebox

I chose Python as a programming language here because, as you could probably guess, there are myriads of libraries in Python for working with music. I found mingus to be the simplest for working with guitar chords and music21 for generating the piano score. It might be possible though to do everything with music21 only, because it's extremely powerful, but I found mixing the two libraries to be easier than implementing the functionality I needed from mingus with music21.

If you are interested in doing anything with Python, be sure to check out Music21, which has a lot of cool features — including doing twelve-tone matrix stuff like this — built-in.

http://web.mit.edu/music21/ Music21 documentation.

It's definitely a viable possiblity, and there's quite a few companeis already doing it. If you want to explore doing it yourself, I'd check out https://web.mit.edu/music21/ and build some basic models using LSTM etc. to have some fun using open source MIDI data sets like https://magenta.tensorflow.org/datasets/maestro .

A little manual, but music21[0] can do the analysis!

[0]: http://web.mit.edu/music21/

There's no universally accepted term for the latter concept that I know of. For example, the music21 music processing library calls the general class that includes notes, chords and rests Music21Objects, and you'd think they'd call them something better if they could.

This Python toolkit has support for analyzing folk tunes. https://web.mit.edu/music21/

(define-public music21 (package ... (build-system python-build-system) (arguments `(#:phases (modify-phases %standard-phases (replace 'check (lambda* (#:key tests? inputs outputs #:allow-other-keys) (when tests? (add-installed-pythonpath inputs outputs) ;; See: https://github.com/cuthbertLab/music21/issues/1164 (invoke "python" "-m" "music21.stream.tests"))))))) ...))

openai/jukebox: Music Generation

https://github.com/openai/jukebox The demo code is there.

I didn't even know about it until I heard Sam Altman casually mention it in an interview, I was expecting some basic tunes generator, but this is so amazing! I mean yeah the voices are not clear, it's muffled, but look at how far have image models progressed, if you applied the same amount of collaborative effort here, the results could be amazing! ElevenLabs showed how good and clear can AI-created voices sound. The only reason I can think of is that the Jukebox code is under view license only.

OpenAI had this figured out 3 years ago: https://openai.com/blog/jukebox/ . You could then even define your own text. Model is open source too.

They've had jukebox for a few years now, so I'm sure some new model will get released and explode overnight, like what chatGPT did.

That already exists

This model is designed to output raw audio.

However, there are many models which do output midi. That's actually much simpler, and has been done already a few years ago.

I thought OpenAI did this. But then, I might misremember, because their Jukebox actually also seems to produce raw audio (https://openai.com/blog/jukebox/).

However, midi generation is so easy, you even find it in some tutorials: https://www.tensorflow.org/tutorials/audio/music_generation