On Generative/Generated Music

 

6/7/2015

Alas, generative music never took off. For a while it was thought to be the future, where music would randomly compose itself from individual segments stored in a database--essentially a music AI. 

But I've had a change of heart about it, at least from a musical standpoint (and may change again) but there is another use, namely using it to track usage and rights.

Bitcoin may never be used widely as currency, but using the blockchain might be interesting to mine and track music metadata. How that gets implemented is still hugely complex.

It could work like this: Individual segments from music would be saved as individual files, in increments ranging from note durations to various phrase lengths (in measures), and would be meta-tagged using the Music Ontology schema.

The blockchain would prove ownership of the (parent) master copy, along with an accompanying waveform. Just as a photographer can prove ownership through sites like Ascribe, a songwriter could assign or assert rights in a composition or its derivations.

[Post-script 6/28/2015: Amazon's new Kindle Edition Normalized Page Count royalty payments by pages read have interesting implications for the form of the content. What if in fact all intellectual property was similarly micro-monetized? If musicians were paid by the bar or note value, art would begin to be shaped by jiggering the metrics to be paid more: more/less notes/durations, etc. This is another example of the medium creating messages].

If we know we are paying by how many bars we listen to or how many pages we read, it makes it a game. In the rush to make things remunerative, we remove the intrinsic aspects of creativity.

[6/7/2024: Generative music ultimately did take off with AI-generated music. But “generated” and “generative” are two different worlds. I doubt anyone that has used the new generation tools would find the old Sseyo Koan program very interesting].