Making a moonshine server

[briangavin]

I am fairly new to python. I have a looked at a number of different whisper ASR servers. I feel like most people use Whisper for transcribing recorded video and audio not live speech. For moonshine, I want to do it for live speech. I have made a web socket server that also does the speech transcribing of the microphone and runs pretty good on a raspberry pi 4 and really good on a raspberry pi 5. The latency is so much better on moonshine. Most of the whisper servers I have seen do not manage the microphone the server is passed raw audio. I feel like project like home assistant might also benefit from using moonshine instead of whisper.

What are peoples thoughts on what a moonshine server should do? Is there an API that should be used?