For my talk at Deconstruct 2018, I made a context-free grammar to transform plain melodies into bluegrass banjo style. Here's an example:
The first 25 minutes of my talk is faux live coding this banjo grammar.
The last 10 minutes is my personal story, “crying in my banjo guru’s bathroom.” It's about how you can learn hard things, like how to be fluent in a language, whether it’s music, code, or speech.
Here's the Jupyter Notebook. You can see the Python code, listen to 22 music clips, and read Markdown explanations step by step.
(The notebook file is a 30 MB download, because it has embedded, uncompressed audio. The notebook should be viewed with nbviewer, because GitHub does not render audio in notebooks.)
If you want to see the code and listen to the music, you don't need to install anything, you can just view the notebook.
If you want to change the code and run it for yourself, please follow these instructions:
- Clone the repository.
- Download & install Anaconda distribution, Python 3. (Anaconda includes Jupyter for notebooks, NLTK for context-free grammars, and more.)
- Download & install these three command line utilities, and add them to your system path. (Binaries are available for Linux, Mac, and Windows.)
- abcm2ps (for ABC → SVG sheet music)
- abc2midi (for ABC → MIDI music)
- Qsynth (includes fluidsynth, for MIDI → WAV audio)
- Download GeneralUser GS v1.471.sf2. (It's a "soundfont", used by fluidsynth.) Put it in the same directory as the repository.
Visit the wiki (a work in progress) for links to learn more about related topics.
I'd love to hear from you! Please contact me at rrherr@gmail.com or on Twitter, @rrherr.