DeepJ: Style-Specific Music Generation | Hacker News Reader

DeepJ: Style-Specific Music Generation

github.com

76 pointslainon18 comments

In terms of melodies I think trainning off lead-sheets is far better than raw midi. The reason being is when humans generate melodies they can anticipate implied harmonic structure. i.e.Chords. You folks might be interested in a leadsheet dataset i'm creating for the purpose of generating melodies: https://github.com/shiehn/chord-melody-dataset

Lead sheets are only suitable for songs though. This work focuses on more complex/polyphonic music.

Ah ok, I really just made an assumption that the goal is to generate convincing harmonic music. In this case after spending a couple years fighting with that problem I've been reaching a conclusion that the chords and melody are absolutely by far the most difficult & 'human' aspect where as counter-point/polyphony are actually very mechanical & really just support the chord/melody structure. Recently I've been having very good results by first generating chords/melody with RNN's then having subsequent processes generate supporting harmonies.

Can you show us some of those good results?

These are results from starting with just a chord progression & stacking ontop of it: http://signalsandsorcery.com About 5 months ago I decided to pivot & start with a melody & chords. I have not yet stood up the system that starts with chords & melody. So the answer is not yet.

I wonder how great composers go about creating a novel melody? Like Mozart or Beatles.

Ya, good question. I doubt there is a single method. I suspect pre 20th century single note melodies were often sung and then following the strict rules of counterpoint harmonies were added. Its seems like pop music is often a chord progression first and then a melody added second.

This project would really benefit from some examples.

Agreed, not really sure what the point of the website is if it's just some sliders with nothing actually playing.

https://github.com/calclavia/DeepJ/tree/icsc/archives/v1

These sound pretty awful, frankly.

Same old story. Short incoherent melodies scattered everywhere with no structure whatsoever.

Next step will be to clone ranking from some well-known music critics which would actually make it work this time. Behavioral cloning is now used heavily in computer vision, haven't seen it used in music generation yet, but I expect it to be groundbreaking as algorithm would be able to discern certain qualities that pure generating sequences won't.

That describes 90% of jazz too.

And 100% of glitch music.

paper: https://arxiv.org/abs/1801.00887

`requirements.txt` has a linux only wheel - any help on getting it running with OSX?

Just install pytorch.

Reply on news.ycombinator.com