This is a rudimentary Node-RED implementation to directly stream audio input from a selected microphone to the Watson Speech to Text API.
At the moment this module is somewhat reusable and should work on windows, mac & linux - however it's only maintained & tested to work on a rasperry pi 3.