How to implement video communication through a browser with speech recognition of one of the interlocutors?

Question

Speech recognition

Hello Toaster audience! Please indicate the path, give parting words, recommendations, tips for implementing video communication through a browser between two users and parallel speech recognition of one of the interlocutors (and converting it to text (Google Speech API or Yandex SpeechKit)) ? in short, if: I need a video link with subtitles)
In the process of excavations on the Internet, I found the following:

• libraries sip.js, jsSIP, PeerJS
  
• gossip about the need to use Stun/Turn servers
  

in general, the documentation on the implementation of video communication is complete.
I need help in choosing a technology, a library, the concept of creating a video call in a browser with simultaneous speech recognition .
How is it possible to separate audio from video, so that it can then be transferred via API to the recognition service

Answer 1

Microsoft had DirectShow and Media Foundation libraries for working with audio and video (interfaces implement a graph), but I don’t remember the network login and documentation of recent years.
Here are the books:
1) Mark D. Pesce. Programming MS DirectShow for digital video and television
2) Turcan, Wasson. Fundamentals of Audio and Video Programming for Games
3) Anton Poligner. Developing MS Media Foundation Applications