A Cartesia MCP for voice workflows that live inside your automations
Connect one API key and Cartesia's Sonic text to speech, Ink transcription, voice cloning, voice localization, speech infill, and usage reporting become workflow steps. An automation can turn a finished draft into narrated audio, a transcript can land in the database next to your tasks, and an AI agent can manage your org's voice library without anyone opening the Playground.
What the Cartesia MCP connector does
The Cartesia connector for Mattering Automations is 13 nodes across the Cartesia REST API: text to speech with the Sonic model family (sonic-3.5 by default), batch speech to text with Ink, voice cloning from an audio sample, voice localization into new languages, speech infill between existing clips, the voice and accent listings that feed every voice picker, and org-level credit and agent usage reporting. One API key authenticates all of it; Mattering sends Cartesia's required version header on every request automatically, so the credential is just the key.
Every node speaks MCP, the open protocol for connecting AI assistants to tools, so Claude, ChatGPT, and any MCP-capable assistant can use them, alongside Matter, the AI assistant built into the platform. An agent can list the voices available to your account, pick one by language and description, generate the audio, and hand the result to the next step of the workflow.
Cartesia publishes its own official MCP server, and it is good at what it is for: giving a desktop assistant like Claude Desktop or Cursor a direct line to Cartesia. This connector is for the other half of the job, where Cartesia is one step among many. On the Mattering canvas the same nodes sit next to every other connector and the platform itself, so a publishing workflow can generate narration and then open a review task, a transcription can write straight into a Mattering database, and a credit check can gate a batch job before it spends anything. Use whichever fits, or both.
The honest framing: this is Cartesia for automation, not for live conversation. Cartesia's streaming transports, the WebSocket and server-sent-events endpoints that power real-time voice agents, are deliberately out of scope for a REST connector, so these nodes generate, transcribe, and manage rather than talk in real time. Of the 13 nodes, 6 read and 7 act, and exactly one is destructive: the voice delete, which permanently removes an owned clone or localization. Generated audio returns as base64 inside the node output under a 5 MB encoded cap, with mp3 as the default container because it holds minutes of speech where uncompressed wav holds seconds. Voice cloning ships with its consent obligation stated on the node: only clone voices you own or have documented consent to clone. The two usage nodes require a Cartesia admin key; every other node works with a standard key.
One connection, three surfaces
- Visual automations. Drag Cartesia nodes onto the canvas and wire them to triggers, conditions, and other apps.
- AI agents, yours or ours. These nodes speak MCP, the open protocol, so Claude, ChatGPT, and any MCP-capable assistant can use them, and Matter, the agent built into the platform, uses them natively.
- The whole platform. Combine with native tasks, projects, and other connectors, no middleware in between.
Every Cartesia node, ready to drop on the canvas
13 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.
Read and query nodes 6
Cartesia: Get API Status
Check that the Cartesia API is up and see the current API version date. This endpoint requires no authentication, so it proves connectivity but not the key; use List Voices to verify the credential.
Cartesia: Get Agent Usage
Get the org's voice agent usage over a time window as day, week or month buckets: billed cents, agent minutes and call counts per bucket. Requires an ADMIN API key (sk_car_admin_ prefix, created at play.cartesia.ai/keys); a standard key returns 403.
Cartesia: Get Credit Usage
Get the org's credit consumption over a time window as day, week or month buckets, optionally filtered to one API key or grouped by capability, model, voice or key. Each row carries its bucket's start_ts, end_ts and credits. Requires an ADMIN API key (sk_car_admin_ prefix, created at play.cartesia.ai/keys); a standard key returns 403. The natural guard node before a big TTS batch.
Cartesia: Get Voice
Get full metadata for one voice: name, tagline, description, gender, language, supported locales and access settings.
Cartesia: List Accents
List every accent Cartesia supports, with its id, display name, language and locale. Accent ids feed the accent fields on the voice clone, localize and update nodes; an accent must match the voice's language.
Cartesia: List Voices
List the voices available to this account: Cartesia library voices plus the org's own cloned and localized voices, with their IDs, languages, genders and descriptions. Filter by search text, ownership, gender or language. This is the free lookup that feeds voice_id into every speech node, and the credential health check for the connector.
Action nodes 7
Cartesia: Clone Voice from URL
Clone a new voice from a short audio sample fetched from a public HTTPS URL (flac, mp3, ogg, wav or webm; this node downloads up to 20 MB and uploads it as the clip). Returns the new voice's metadata including its id, ready for the speech nodes; reversible with Delete Voice. Only clone voices you own or have documented consent to clone; you are responsible for having the rights to the sample. Direct file upload is not supported on the canvas, so host the clip at a URL first.
Cartesia: Delete Voice
Permanently delete an owned voice (a clone or localization) from this account. The voice cannot be recovered; Cartesia library voices cannot be deleted. Returns an empty 204 on success.
Cartesia: Infill Speech Between Clips
Generate speech that splices naturally between existing audio clips: give a left and/or right context clip by URL (at least one; this node downloads up to 20 MB each) plus the transcript to infill, and Cartesia matches prosody across the seam. Returns the generated audio as base64 (data.audio_base64) subject to the 5 MB encoded cap. Use the same voice and output format as the surrounding clips.
Cartesia: Localize Voice
Create a new voice that speaks a different language with the same vocal character as an existing voice. Returns the new voice's metadata including its id; reversible with Delete Voice.
Cartesia: Text to Speech
Convert text into speech with a Cartesia Sonic voice. Returns the audio as base64 (data.audio_base64) with its content type and byte size; responses over the 5 MB encoded cap fail with a clear error, so use mp3 at a lower bit rate or shorter text for long passages. Billed in credits per character. Timestamped output exists only on Cartesia's WebSocket transport, which is out of scope for this connector.
Cartesia: Transcribe Audio from URL
Transcribe an audio file with Cartesia's Ink batch speech-to-text by pointing at a public HTTPS URL (this node downloads up to 20 MB and uploads it for transcription; long files are chunked by Cartesia automatically). Returns the transcript text, detected language, duration and optional per-word timestamps. Billed by audio duration. Direct file upload and realtime streaming are not supported on the canvas, so host the file at a URL first.
Cartesia: Update Voice
Update an owned voice's metadata: name, tagline, description, gender, accent or access. Omitted fields keep their current values; read them first with Get Voice if you need to restore later. Only voices this org owns (clones and localizations) can be updated.
Connecting Cartesia takes one credential
Cartesia authenticates API requests with an API key created in the Cartesia Playground at play.cartesia.ai under API Keys. Standard keys are prefixed sk_car_ and cover everything except the two usage reporting nodes, which need an admin key (prefixed sk_car_admin_, created at play.cartesia.ai/keys/admin). Cartesia also requires a Cartesia-Version date header on every request; Mattering sends both the X-API-Key header and the version header for you, so the only thing you configure is the key itself.
Sign in to the Cartesia Playground at play.cartesia.ai, open API Keys, and create a key. Create an admin key at play.cartesia.ai/keys/admin instead if your workflows will use the credit or agent usage nodes; a standard key covers everything else.
Copy the key. Keys are prefixed sk_car_ (admin keys sk_car_admin_).
In Mattering Automations, open Connectors, click Add Connector, choose Cartesia, and create a credential with the API key.
Click Test Connection. Mattering calls Cartesia's voices list, which works on every key: a voice list back proves the credential, an error means the key is wrong or revoked. The unauthenticated API status endpoint is deliberately not used here, because it succeeds even with a bad key.
All 13 Cartesia nodes appear in the canvas palette. Run Cartesia: List Voices first: it supplies the voice IDs that every speech node asks for.
Common uses for the Cartesia MCP
Turn finished writing into audio
The core Cartesia automation: a workflow takes text from wherever it already lives, a database row, a task description, an inbound webhook, and Cartesia: Text to Speech converts it with a Sonic voice. The audio comes back as base64 in the node output with its content type and byte size, mp3 by default because mp3 at the server default bit rate holds minutes of speech under the 5 MB cap. Billing is in credits per character, so the credit usage node makes a natural guard step before a long read.
Transcripts that land next to the work
Point Cartesia: Transcribe Audio from URL at a recording hosted anywhere public over HTTPS and Ink returns the transcript text, detected language, duration, and optional per-word timestamps. Long files are chunked by Cartesia automatically. A scheduled sweep can transcribe the week's recordings into a Mattering database, or a webhook can transcribe an upload and open a task with the transcript attached, no one pasting audio into a tool by hand.
A voice library your org actually manages
Cartesia: Clone Voice from URL creates a new voice from a short public audio sample, for a voice you own or have documented consent to clone, and returns its ID ready for the speech nodes. Cartesia: Localize Voice makes that same vocal character speak another language, Cartesia: Update Voice keeps names, taglines, and accents current, and the destructive-flagged Cartesia: Delete Voice retires clones and localizations cleanly. An agent can run the whole lifecycle, with the delete behind an approval step.
Fix one sentence without regenerating the whole read
Cartesia: Infill Speech Between Clips takes a left and right context clip by URL plus the corrected transcript, and generates speech that matches prosody across the seam. That turns a one-word fix in an hour of narration into a single node run instead of a full regeneration, which matters when generation bills per character. Use the same voice and output format as the surrounding clips so the splice is invisible.
Credit guardrails before batch jobs
Cartesia: Get Credit Usage returns the org's credit consumption in day, week, or month buckets, optionally grouped by model, voice, or API key, and Cartesia: Get Agent Usage does the same for voice agent minutes and calls. Both need an admin key, stated plainly on the nodes. A batch narration workflow can check the month's consumption first and stop, or open a task for a human decision, before a big run spends the remaining credits.
Example: A weekly digest becomes audio: check credits, generate, hand off for review
Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.
Trigger: a schedule fires weekly, or an inbound webhook delivers the finished digest text from your publishing workflow.
Cartesia: Get Credit Usage reads the month's credit consumption with an admin key. A condition step stops the run, or routes to a human, if the budget is nearly spent.
Cartesia: List Voices supplies the voice ID, filtered by language and search text, the same lookup that powers the voice pickers on every speech node.
Cartesia: Text to Speech converts the digest with a Sonic voice, returning mp3 audio as base64 with its content type and byte size.
The workflow opens a review task in Mattering Productivity with the run's output linked, so a person signs off on the audio before it goes anywhere public.
Cartesia MCP questions, answered
What Cartesia operations are supported?
13 nodes on the Cartesia REST API: 6 query nodes and 7 action nodes covering text to speech with the Sonic family, batch speech to text with Ink, voice cloning from a URL-hosted sample, voice localization, speech infill between clips, voice management (list, get, update, delete) and the accents list, credit and agent usage reporting, and an API status check. Deliberately not shipped: all streaming transports (WebSocket and server-sent events), and the datasets fine-tuning surface, which is upload-centric and not automation-shaped.
Does this work with Claude, ChatGPT, or other AI assistants?
Yes. Mattering exposes these nodes over MCP, an open protocol for connecting AI assistants to tools, so any MCP-capable assistant, including Claude and ChatGPT, can connect to your workspace and use them. Matter, the assistant built into the platform, uses the same nodes natively.
Cartesia publishes an official MCP server. Why use this connector?
Cartesia's official MCP server is a genuine option when all you want is an assistant talking to Cartesia alone, and it runs locally alongside tools like Claude Desktop and Cursor. This connector earns its place when Cartesia is one step in a larger workflow: the same canvas holds your other connectors and the Mattering platform itself, so a transcription can write into a database, a narration job can wait behind a credit check or an approval task, and the credential lives in one place with the rest of your connector secrets. The same nodes serve visual automations and AI agents alike. Use whichever fits, or both.
Can this power a real-time voice agent or phone call?
No, and the page says so on purpose. Real-time conversation runs on Cartesia's streaming transports, the WebSocket and server-sent-events endpoints, which are out of scope for a REST connector. What these nodes do is voice automation: generate finished audio from text, transcribe recorded audio, manage voices, and report usage, including agent usage from Cartesia's voice agent product. If you are building live voice, build it on Cartesia's own SDKs and use this connector for the workflow around it.
How does the generated audio come back?
As base64 inside the node output, with its content type and byte size, capped at 5 MB encoded. The text to speech node defaults to the mp3 container because mp3 holds minutes of speech under that cap, where uncompressed wav at 44.1 kHz reaches it in under a minute of audio. A response over the cap fails with a clear error advising mp3 at a lower bit rate or shorter text, rather than silently truncating.
Can an automation clone a voice?
Yes, with the obligation stated where it belongs. Cartesia: Clone Voice from URL creates a voice from a short audio sample fetched from a public HTTPS URL, and the node's own description carries the consent requirement: only clone voices you own or have documented consent to clone. You are responsible for having the rights to the sample. Cloning here is for your own voice assets, brand narrators, localized variants of a consented voice, never for imitating third parties. The clone is reversible with the voice delete node.
Do I need a paid Cartesia plan?
Cartesia's pricing page lists a Free tier with monthly credits that cover Text to Speech and Speech to Text, alongside Pro, Startup, Scale, and Enterprise plans, and API keys are created in the Cartesia Playground. Two things to check against your use: the pricing page lists a commercial use license from the Pro plan up, and it does not spell out API access tier by tier, so confirm the tier that fits how you will use the audio on Cartesia's pricing page. Text to speech bills in credits per character and transcription by audio duration.
Why do the usage nodes return a 403 when everything else works?
The credit and agent usage endpoints require an admin API key, prefixed sk_car_admin_ and created at play.cartesia.ai/keys/admin, and a standard sk_car_ key gets a 403 there by design. Every other node works with a standard key. If your workflows need the usage guardrail, create an admin key and store it as the connector credential, or keep a separate admin credential just for those nodes.
Put Cartesia to work inside your workspace
Connect once, then use every node in visual automations and AI agent workflows across the Mattering platform.
All product names, logos, and brands are property of their respective owners and are used for identification purposes only. Use of these names, logos, and brands does not imply endorsement or affiliation. Cartesia is a product of its respective owner; Mattering is an independent platform.