Project to analyse conversational emotional levels

hi there, I want to use a rpi zero W with a microphone in order to stream the audio recorded to be near-real time analyzed in a server to output a CSV file with the emotional levels, every x seconds or cut audio files.

I have previous exp with rpi B with dietpi and a grovepi hat with sensors

I guess I need:

  • rpi zero W, microSD, USB keyboard, charger… which waterproof case?

  • which software for streaming audio to the server?

  • in wav or mp3?

  • which microphone for good quality recording should I use (USB or 3.5mm mic)?

  • and golden question, which software to process in server the emotional levels (I guess some AI, maybe any open source?)

I know these are many questions, just wondering your thoughts

Thanks in advance