The goal of the Sound Signal Synthesizer (SSS) was to write a program that could synthesize a short piece of music that sounds “good” (to me at least). It features a bigram model that uses pieces of music it is previously trained on to generate new music, as well as a constrained probabilistic generation model that randomly generates new notes within a few custom constraints.

The resulting music is not necessarily what I would call “good,” but I am happy with what I was able to produce given the resources I had. Please see the presentation for more details!