Demo using the test.wav file supplied with the vosk distribution. To use this (unless your machine is set up exactly like mine), pass in a path to where you have unpacked your voice model. Models can be downloaded from here.

A listener function using vosk as the sound-to-text stage. Listen on this stream and interpret the input using the model indicated by the :model-path value in this config. If no config is supplied, will look in the environment for an environment variable VOSK_MODEL.

Heavily cribbed from the DecoderDemo given here.