Stochastis

Convert Python code into music to hear algorithms

About

Pronounced "stoʊ-kæst-ɪs" ("stow" + "cast" + "is" like "miss"), Stochastis was a small experiment in figuring out if it would be possible to hear what a given algorithm sounds like. However, the goal was accomplished in a much more general fashion. Instead of only algorithms, Stochastis converts each function in a Python module to its corresponding bytecode which it then converts to MIDI notes along a predefined scale. It then generates a MIDI file from them or plays them upon generation.

Installation

$ pip install stochastis
$ pip install git+https://github.com/Pebaz/stochastis

Usage

$ stochastis foo.mid                     # Play from file
$ stochastis some_file.py                # Generate and play from file
$ stochastis some_file.py --out foo.mid  # Generate only

Conclusion

What was accomplished in this experiment is just a small inkling of the amount of potential this has for pleasing music being generated from a given codebase.

Some ideas for extensions include:

• Identifying patterns in bytecode to produce familiar stanzas
• Performing analysis on the bytecode of functions to identify patterns so that pre-built sequences of notes can be grafted into the output
• Clearer rules for what is generated based on what is actually written in Python code
• Integration with better sounding synthesizers
• Layered tracks with multiple intstruments playing at once (drums, etc.)
• Converting program control flow into sections of a given song