How Google Meet Implements Audio using Mix-Minus with WebRTC

Google Meet, a popular video conferencing tool, implements a sophisticated audio management technique known as mix-minus using WebRTC to enhance call quality. This method involves sending only the three loudest participants' audio through separate tracks, ensuring that feedback and echo are minimized by subtracting the speaker's audio input from their output. Through careful observation using tools like Chrome’s WebRTC-internals, it was discovered that Google Meet dynamically remaps participants to different audio tracks when multiple speakers share the same track, indicating that each track can only carry one participant's audio at a time. This remapping ensures that the audio delivered to each participant remains clean and clear, without mixing multiple participants' audio into a single stream. Additionally, the platform's architecture likely involves the use of a Detector and Audio Engine to manage and deliver the audio streams effectively, providing each presenter with a mix-minus audio stream based on current loudness metrics.