In the past couple of days I created a short sound piece using ChatGPT. ChatGPT works with text, not sound, so my approach was to ask it to create a program that synthesizes sound of a certain description. Similar could be done, I imagine, with prerecorded sounds, like I could ask it to create a program that combines prerecorded sounds in a certain way, but I haven't tried that yet. The resulting synthesized sound is simpler than I had originally intended, because it turned out that there is quite a limitation on how complex a program ChatGPT can create before going "off the rails" and losing the thread. After the conversation I asked it to summarize the last successful result into a new prompt. The result is this 1-minute piece to which I gave the title Etude on Sound Blocks with Normal Distribution.

Etude on Sound Blocks with Normal Distribution

Computer music/noise with Java or C code. Also songs, instrumental music, improvisation.

Etude on Sound Blocks with Normal Distribution

Andreja Andric