ElevenLabs Realtime

Deliver ultra-low latency voice experiences (TTS and STT) with Flash v2.5 and Scribe v2 Realtime, enabling responsive, human-like AI agents.

ElevenLabs Realtime technology provides the speed required for true conversational AI: no more lag. Our flagship real-time Text-to-Speech model, Flash v2.5, delivers speech synthesis with ultra-low 75ms latency across 32 languages, perfect for interactive applications and voice agents. Complementing this is Scribe v2 Realtime, our Speech-to-Text model, which achieves state-of-the-art accuracy in over 90 languages with just 150ms of latency (down to 30-80ms in optimized setups). These models are engineered for performance, ensuring your AI agents and live transcription services operate at the speed of human conversation.

https://elevenlabs.io/docs/text-to-speech