An experiment: is it possible to work out the note being played on a piano with the Web Audio API?

1. git clone https://github.com/davidgilbertson/sight-reader
2. npm install
3. npm build
4. node index
5. http://localhost:8080

To run in dev mode, run npm run dev.

The result. No, it's not really possible. It gets the note right 80% of the time if it's loud, but often jumps about as the note trails off.

It gets three of the four samples when played directly into the audioContext but only the two middles ones when played out the through the speakers, in through a mic and captured with navigator.getUserMedia

Weird build

Also, this is an experiment in packaging without packaging. It's dumb, but...

• Each file is an IIFE that puts its constants/class on a global object called SP_APP, that's the equivalent of 'exporting' the module
• Classes are then accessed off this object like const {Piano} = SP_APP - the equivalent to import Piano from './Piano'.
• There is no actual importing/exporting, so the site is 'built' by just concatenating all files.
• So when main.js lands in the browser and is parsed, the SP_APP object gets populated with constants and classes and utils and whatever else.
• Because there's no importing, it means no npm packages allowed.
• In the HTML file, window.SP_APP = {} is created, then the single script is loaded/executed, then SP_APP.start() is called, so I know that all classes are defined and ready to go before I start executing any code (e.g. I can instantiate one class in the constructor of another class because that will have been created and added to SP_APP at that point)