For our project, we created an orchestra of AI agents that are capable of listening to music and recreate a song in Minecraft by scheduling each one to hit the appropriate Minecraft Noteblock on time in an optimal manner. Our input is an audio file (such as .wav or .mp3) and the output will be the sound genereated from the noteblocks the AI hits, as well as a log that tells what the AI hit with the timestamp.
For this project we will treat this problem as a constraint satisfcation/optimization problem. We will process the audio input into a csv file that contains the frequencies of notes played at time t. The new frequency list allows us to treat this problem in a discrete time domain and sets up this optimization scheduling problem as a constraint satisfaction problem.
The audio generated by the agents of the simulation can be recorded and compared to that of the original song by frequency and amplitude values.
We can compare the audio generated by the output of the agents of the simulation and compare that to the original audio of the song being modeled and compare the quality of the recreation.
Our group time is Apr. 26 at 10:45pm