How can I access voice/video files in realtime?

[romeronathan]

Basically, my use case is when a user speaks during a stream, I'm trying to pipe that audio into a speech to text model so they have real time transcription. Can somebody point me in the right direction to accomplish this using AMS? 

[Connessione]

I think ant media has filter plugins to help you access raw data on serverside. But for your case i would first explore the possibility of using mic input from the browser into third party speech recognition engines via their sdks.

For example Google speech to text SDK has client side example of capturing mic input using SDK methods.