It sounds great although I think would be more impressive if he was able to use machine learning to produce a heavy metal record lol. Making great music with math and synthesis is easy what is more interesting is to develop a GPT3 for music where you could type a description of what it should sound like and get an accurate representation of your idea out of the other end of the software without knowing anything but non-musicians description of a band or piece of music.
Very interesting and the music is more than decent. What was the fail rate? Did you ask friends for opinions on half made tracks? You have some ear for musical finesse - you do seem to know something about the process of making music.
It takes me a couple of hours to go from 0 to a half-baked ~2m track. Using this process, I ended up creating more than 100 tracks, from which I selected 8 for further development and mastering. 6 of them ended up in the album.
- Did you ask friends for opinions on half made tracks?
Yes, I have some friends who work in the music industry, and they served as my test bench during the process.
- You have some ear for musical finesse - you do seem to know something about the process of making music.
Thanks! I been actively listening to music for a couple of decades now, and got decent at the task of identifying why I like a track or not. Just tried create a tool that allowed me to leverage that experience :)