When building a WebRTC voice call, you can't pass a remote MediaStream to the SpeechRecognition API. Here are three workable approaches — remote-side recognition, server-side processing, and AudioContext — plus iOS-specific implementation strategies.
