# Cartesia

**Canonical:** https://apis.io/providers/cartesia-ai/  
**Website:** https://cartesia.ai/  
**APIs profiled:** 21

Cartesia builds real-time voice AI - the Sonic family of text-to-speech models, Ink speech-to-text models, and a Voice Agents platform for building and deploying telephone and web voice agents. The core generation surface is exposed both as REST (bytes and Server-Sent Events) and as a low-latency, bidirectional WebSocket protocol at wss://api.cartesia.ai for streaming TTS and STT with multiplexed contexts, word/phoneme timestamps, and mid-stream flush and cancel.

## Kin Score — 43.0 / 100 (developing)

Scored 2026-08-24 under rubric 0.13.0. Trend: flat (-0.3 from 43.3).

| Facet | Score |
|---|---|
| Discoverability | 57.4 |
| Contract Quality | 62.9 |
| Governance | 11.4 |
| Contract Governance | 11.4 |
| Operational Transparency | 34.2 |
| Developer Ergonomics | 28.6 |
| Commercial Clarity | 50.0 |
| Access Clarity | 50.0 |

## Agent readiness — 22.6 (agent-aware)

| Dimension | Value |
|---|---|
| Spec Presence | yes |
| Agentic Access | derived |
| Reversibility Documented | documented |
| MCP Server | no |
| Auth Clarity | bearer |
| Idempotency | no |
| Error Semantics | no |
| OpenAPI Examples | no |
| Rate Limit Signal | documented |
| Event Surface Described | derived |
| Agent Skills | no |
| Well Known Catalog | no |
| Consent Identity | no |
| Agent Card | no |
| Dry Run Mode | no |
| Delegated Identity | no |
| Protected Resource Metadata | no |
| Dynamic Client Registration | no |
| Agentic Commerce | no |

## Access

Freemium · Self-serve signup — onboarding: self-serve, pricing: freemium, trial: no (confidence: high).

## APIs (21)

- **Cartesia TTS WebSocket API** — Bidirectional, multiplexed WebSocket at wss://api.cartesia.ai/tts/websocket for realtime speech generation. Clients send JSON generation requests keyed by a context_id (continui...
- **Cartesia STT WebSocket API** — Realtime transcription over WebSocket at wss://api.cartesia.ai/stt/websocket. Clients stream raw binary audio in 100ms chunks and issue finalize/close text commands; the server ...
- **Cartesia Agents API** — Voice agent configuration and public templates.
- **Cartesia API Keys API** — Standard API key metadata.
- **Cartesia Auth API** — Short-lived, scoped access token generation.
- **Cartesia Calls API** — Outbound call placement, batching, and history.
- **Cartesia Datasets API** — Audio/text collections used to build custom fine-tunes and voices.
- **Cartesia Deployments API** — Agent deployment inspection.
- **Cartesia Infill API** — Generate audio that smoothly connects two existing audio segments.
- **Cartesia Knowledge Base API** — Documents and folders used by agents.
- **Cartesia Metrics API** — LLM-as-a-judge quality metrics for agents.
- **Cartesia Phone Numbers API** — Telephony number provisioning and import.
- **Cartesia Pronunciation Dictionaries API** — Custom pronunciation overrides referenced by TTS requests.
- **Cartesia Providers API** — Linked telephony provider accounts.
- **Cartesia Speech-to-Text API** — Batch transcription of an audio file of any length.
- **Cartesia Status API** — API health and version.
- **Cartesia Text-to-Speech API** — Single-shot and streamed speech generation over plain HTTP.
- **Cartesia Usage API** — Credit and agent usage reporting.
- **Cartesia Voice Changer API** — Transform speech into a different target voice.
- **Cartesia Voices API** — Voice catalog management and voice cloning.
- **Cartesia Webhooks API** — Webhook endpoint registration.

## Agentic access (1)

- **Cartesia Ai Agentic Access** — 86 operations · 48 acting

## Security (2)

- **Cartesia Ai Authentication** — http · 1 scheme
- **Cartesia Ai Domain Security** — TLSv1.3 · HSTS · DMARC

## Plans (1)

- **Cartesia Ai Plans Pricing**

## Tags

Artificial Intelligence, Voice AI, Text-to-Speech, Speech-to-Text, Real-Time, WebSocket, Voice Cloning, Voice Agents

---

Profiled by [API Evangelist](https://apievangelist.com) and published on [APIs.io](https://apis.io/providers/cartesia-ai/). Scores are computed from the provider's own public artifacts under a published rubric.