I am Brendon, a self-employed (digital nomad) programmer. I ran an experiment which attempted to automatically write a melody, based on the votes of anonymous internet visitors.
Now, another experiment has begun to write lyrics for the same song, one word at a time:
Starting from 2 given notes, the voter was asked which sequence sounded best, when an extra pitch was added to the end of the sequence:
[Note 1] [Note 2] [A/B/C/D/E/F/G] <- Which sequence sounds best?
The winning vote generated a new note and the crowd then voted on a longer sequence:
[Note 1] [Note 2] [Note 3] [A/B/C/D/E/F/G] <- Which sequence sounds best?
This process continued until the sequence became the length of an entire melody.
My theory was that if this system was extracting and expressing knowledge about what the majority enjoy listening to (at the most granular level)...the crowd should be able to generate their own song (which they also enjoy listening to). So the experiment began.
Anyway, the melody is now complete:
You can also hear some of the remixes here:
and vote for the next word of the lyrics:
Here for the next few hours, to answer any questions you have about the project.
You can follow the project on twitter @crowd_sound