Audio Models

Dograh is an open-source voice-agent builder launched August 10, 2026, featuring visual flow design and telephony integration. It provides a self-hosted alternative to commercial voice-agent platforms.

Eleven v4 is ElevenLabs' latest text-to-speech model family, launched September 28, 2026. It includes Eleven v4 (expressive, high-quality speech synthesis) and Eleven v4 Turbo (lower-latency variant optimized for real-time conversational agents). Both are accessible via the ElevenLabs API.

Eleven v4 Turbo is ElevenLabs' real-time, low-latency speech synthesis model released September 28, 2026. Designed for live applications requiring minimal delay, it achieves a median latency of approximately 100 milliseconds. Available through a Text to Dialogue WebSocket API, it complements the quality-focused Eleven v4 model for use cases where speed is prioritized over maximum audio fidelity.