The FMAR Python Audio Transcriber package is deprecated and has been replaced with a different system using a direct implementation of whisper in Unity which does not require running a separate python instance.

The FMAR Python Audio Transcriber provides a proof-of-concept for how audio can be transcribed with a whisper server running on a separate python instance. This demo tests sending audio data to a python server, receiving text responses of the transcribed data, and then attempts to interpret that data into meaningful commands. All of this is done in real-time, or as close to it as possible. Testing of this system made clear that whisper is not reliable for detecting and parsing command data, as it struggles with consistency of output as well as has a hard time with short audio inputs. However, the system is generally good at providing transcripts of longer audio streams.