teams-meeting-audio

To get AI to join a web meeting and talk in real-time, use a script that sets up PulseAudio virtual devices and bridges OpenAI Realtime speech-to-text with ElevenLabs text-to-speech for platforms like Teams and Zoom.

Setting up real-time AI audio for browser meetings requires an Ubuntu environment with PulseAudio, Node.js, OpenAI Realtime API, ElevenLabs TTS, and an optional Exa API for web search.

You can control the AI during Teams or Zoom meetings using voice commands triggered by saying the AI's name to execute actions like leave, mute, pause, and resume.

During a Google Meet session, the AI fetches information and takes notes by performing on-the-fly Exa web searches and logging the speech-to-text transcript for later task review.

No manual PulseAudio configuration is required, as a single command automatically sets up the virtual audio devices and launches the meeting bridge.

The best way to automate meeting participation across web platforms is using a real-time audio bridge that integrates speech-to-text and text-to-speech, allowing a single AI agent to actively listen and respond.