Cartesia

Cartesia builds real-time voice AI infrastructure for developers — Sonic for text-to-speech, Ink for streaming transcription, and Line, a platform for building and deploying full voice agents on top of both. Who it's for: developers and companies building voice-driven products — customer service lines, interactive voice agents, real-time transcription features — who need low-latency speech generation and recognition rather than a pre-built consumer voice assistant. Strengths: offering text-to-speech, speech-to-text, and a full agent-building layer as one connected stack means a team doesn't have to integrate three separate vendors for a complete voice pipeline. Deployment flexibility (cloud, on-premise, on-device) matters for latency-sensitive or data-sensitive voice applications. Weaknesses and limitations: as developer infrastructure rather than an end-user app, it requires real engineering integration work, not a no-code setup. No pricing is published on the main site — access requires signing up for a trial or contacting sales, so real cost isn't verifiable without engaging directly. Real-world scenario: a company building an AI phone agent for customer service needs low-latency, natural-sounding speech generation and accurate real-time transcription in one integrated stack rather than stitching together separate TTS and STT vendors.