Yes, it’s definitely possible. Since ElevenLabs doesn’t stream viseme data yet, we can generate real-time visemes by mapping phonemes from the audio stream (via a lightweight phoneme extractor) and sync them with LiveKit’s media tracks for medium–high fidelity lip-sync. I can help design and implement a custom LiveKit viseme plugin that works seamlessly with ElevenLabs in real time. @acceptable-leather-47181