According to Wikipedia, AI (artificial intelligence) is huge area for innovations in researching stock marketing, medical sciences, robot manipulations, mathematical predictions. Art products are not trivial ways to use that powerful tool, but context is getting better every day. «Computer brains» applications appear in music theory, composition and even sound processing.
Machine learning algorithms, counting capabilities of modern computers provide an opportunity to process immense data arrays and accumulate experience using neural networks.
First attempt to create algorithmic music took place about 60 years ago in Soviet Union. That composing was built using the «Ural» monster computer.
Implementations of AI-generated music are varied. Some approaches build on combining prepared patterns. Some algorithms create absolutely unique note sequences. We can find interesting creative solutions, all of them contain special strengths. For example, well-known Amper Music able to generate musical fragments according of genres and even moods. Impressive another project AIVA is recognized by music community as a powerful improved compiler of classical famous composers. A project is based on deep learning solutions with extremely complicated detailed architecture. Next system named Melomics use unobvious bionic methods.
Artificial intelligence in music is immense area for experiments. We are interested, we make attempts, but nowadays our achievements are not too impressive.
I have seen few projects about AI music. Programmers are looking for personal way in creation something worthy. Some of them struggle with trivial forms, some of them dig into musical theory. I doubt their methods. Firstly, it's too simple. Secondly, it's too sophisticated.
So we have got a trash in first case. And we have very specific complex products, full of academic musical phrases, dissonant accords in second case. It can be carefully calculated, but it's too hard for ordinary listener.
Submitted by me AI music generator is simple and no comparison with aforenamed high-budget and resource-intensive projects. But it based on a strict musical theory and could be something much greater in perspective.
So I present my own vision of artificial music generation. I hope it well-balanced as first experience. By the way, this product is not finished. If I get a good feedback, I will continue.
Do you like AI-generated music? Is my project good? Share your opinion, send me feedback.