Hi there,
you need to flash the specialized Edge Impulse audio firmware onto the module and then use the Edge Impulse CLI (Command Line Interface) for real-time inference via a connected computer
There is no direct Arduino example for real-time audio inference using just the Grove AI module itself as the main processor for the audio data.
There is keyword spiting…model though
HTH
GL 
PJ 