companies
Cartesia
Builds real-time voice AI models — including the Sonic family — using state space model architectures instead of transformers for low-latency speech.

Cartesia was founded in 2023 by Stanford AI Lab researchers Karan Goel, Albert Gu, Arjun Desai, and Brandon Yang, whose earlier academic work on state space models (SSMs) offered an alternative to the transformer architecture behind most modern voice AI. The team's pitch is that SSMs handle long audio context more efficiently than transformers, enabling much lower latency for real-time conversation. Its flagship product, Sonic, is a text-to-speech and voice API built on that architecture, aimed at companies building voice agents, IVR replacements, and conversational products where response speed changes how natural a call feels. Sonic has gone through several versions, with Sonic-3 adding expressive delivery like laughter and tonal shifts on top of further latency cuts. Cartesia raised a $64 million Series A in March 2025 led by Kleiner Perkins, followed by a $100 million Series B in October 2025 that added NVIDIA as an investor, and counts customers including Decagon and Cresta among enterprises building voice agents on its API.
- Founded
- 2023
- Headquarters
- San Francisco, United States
- Sector
- lab
Tools
- Sonic
Sonic is Cartesia's text-to-speech model for real-time voice applications, with voice cloning, localization, and multilingual speech.