Here is what my architecture looks like:

(fetch video stream from webcam) => (fetch each frame in stream and convert to byte array) => ( send bytes to webserver using websocket) => (broadcast bytes to online subscribers on different browsers and devices).

I want all of these processes to happen offline so that rules out webrtc since it requires sturn and turn signaling servers and I don't think it provides me the capacity to access video and audio frames so as to generate byte arrays from them for further processing...

I do not want to use webrtc. I have already tried using mediarecorder it works well but there is quite some few milliseconds time lapse which is not ideal for my application and the overall processing and conversions of video frames to byte array just eats up my entire processor space to 100%. Is there a way to access video and audio frames from mediaDevices.getUserMedia({video: true,audio: true}) and transmit them to a remote server?

I need an algorithm a library, a technology or a push in the right direction.

I could also provide source codes if requested.