Am I the only one that’s tired of coming up with my own ideas for music? I can’t be: Writing music is a chore. Every composer knows this, but aren’t willing to admit, because of how much time and money they have spent learning how to do it, despite not enjoying it (sunk-cost fallacy).
To save the time and money of the next generation of composers, and to put Steinberg even more ahead of the curve in the music production and notation market, Steinberg should be collaborating with OpenAI to create DoricoGPT: An AI model that composes, arranges and engraves hours of music in just a few seconds.
“Why?”, you may ask, and “but I love making music”. Well, first off, why not? Robots are quickly replacing everything already. So why not get the upper hand and make the robots work for us instead of the other way around? No more paying ghostwriters to compose your music, and all the legal complications that come with it. Second, no you don’t. You don’t like making music. That is just a trick that the human brain is playing with you, to help cope with your bad decisions that you didn’t like. Implying that replacing the human brain with a digital brain will probably be necessary for all of us in the near future.
Then the robots could listen to the music they have created as well, which means we would save lots of time not listening to music at all … leaving more time for … (golf, Netflix, reading…)
ChatGPT is a language prediction model. Language only works in one dimension (linear), music has multiple dimensions which all depend on each other (linear, vertical (harmony), dynamic/intensity, instrumentation).
A linear prediction model like ChatGPT would need other existing voices as input, which isn’t impossible, but it’s easier to first completely write one voice then the next then the next etc. you also need to allow the AI to go back afterwards and fix mistakes, also something ChatGPT doesn’t seem to be equipped with.
a friend of mine did such a NN to compose Bach Chorals, but it had already a lot of
Musicology programmed in its operational mode.
I have the opposite problem. I come up with new musical ideas easily. It’s the working out of those ideas that gets me down. So in my case I need the AI to do the actual composing. I’ll create the ideas and then turn them over to the AI to do the grunt work.
I was wondering about this recently. I think ultimately human creativity will be still valued: would you watch a film without a human director and composer? But there are some tasks that could be replaced, in the same way symphonic orchestras and live musicians have been replaced mostly by virtual instruments. For instance me, as a composer assistant, could have that part of my income completely wiped. What if an AI can just transcribe MIDI into an orchestral score, with idiomatic writing for each part, fixing errors in orchestration and laying out score and parts.
Also, a film score may likely remain human composed, but what about advertising? one could argue that the same music that was before stock music now will be AI. We’ll see!
