Algorithmic Music
How hard or easy do you suppose, composing algorithmic music is? The obvious answer we would get is easy — and one would recommend to a machine learning architecture like Magenta. But, what’s the issue if we go for that? If you pass Beethoven and Mozart as input, the output will always have a feel of both Beethoven and Mozart and not Chopin!!
Now, if I extend the question slightly by saying how hard or easy do you suppose composing algorithmic music is, without considering machine learning? It looks like even in that case, it’s easy. All you need to have is a library like Music21 and some basic guiding principles.
Music is always about mood. What sort of a mood the musical bit will bring in is all that matters. To achieve a mood, there are a few established parameters-
Mode theory is a bit detailed and is not as straightforward as is mentioned above.
Other words, all you need to do to change the mood is by shifting the notes by x points on the Midi Scale.
Tempo is also not a big issue. Take the case of Polka. Length of notes will look like this. [1, 1, 1.5, 1, 1, 1.5, 1]
Get Balakrishna Ch’s stories in your inbox
Join Medium for free to get updates from this writer.
Rhythm is the number of notes in your key signature, harmony is just something you hardcode.
Loudness is described in music through something called dynamics. If you look at the same in Music21,
I won’t say this will create great music but building upon these basics, may be, we will be in a position to compose some decent music?
Note: While Music21 is for composing traditional music, pyo is another great library for non-traditional music. Even, it’s documentation is a pleasure to read.
