readme is now rst, and included in setup.py
a decoupled part that will prepare audio. It has a modular design based on standard python entrypoints. This allows "kinds" to be registered.