Update, October 28, 2026: the API and MCP server are now available on every plan, including Free. Calls still spend your account's points, so a free account can try both with its starting 60 points.
VocalLab's API and MCP server are both live today. The API lets you generate speech from your own code, against the same voice catalog, your saved clones and your designed voices. MCP lets an AI assistant use that same generation ability directly, without a person opening the app in between.
Both now work on every plan, Free included. This post covers what each one does and what it costs — it isn't a code tutorial, so you won't find endpoint names or sample requests here. Full reference docs live in the in-app API and MCP help articles once you're signed in, and what each plan includes is on the pricing page.
The API, In Plain Terms
The API is for generating speech from your own code instead of pasting text into the app by hand. It reaches the same catalog of stock voices, plus any clones or designed voices already saved on your account, so a script your app sends in comes back read in a voice you've already cast rather than a generic default.
Who gets it
Every plan, Free included. Each call spends your own points.
Rate limits
60 requests per minute, rising to 120 requests per minute on Max.
How it's charged
⌈characters ÷ 15⌉ points per generation — a deliberate premium over the app's ÷17 rate.
| In the app | Through the API | |
|---|---|---|
| Pricing | ⌈characters ÷ 17⌉ points | ⌈characters ÷ 15⌉ points |
| Available on | Every plan | Every plan |
| Rate limit | No request cap | 60/min, 120/min on Max |
Said upfront, on purpose: The API's ÷15 rate costs more per character than generating the same text in the app's ÷17. That's intentional — it's the cost of programmatic access — and it's better to know that going in than to find it on an invoice.
What You Can Reach Through It
- The full voice catalog — the same stock voices you can browse and audition in the app.
- Your saved clones, so a script sent through the API comes back in your own cloned voice.
- Your designed voices, cast the same way you'd cast them from the editor.
What It Doesn't Do
It's worth being direct about the boundaries, because assuming more than what's there is how a first integration goes wrong. There's no streaming or websocket connection: a request returns a finished audio file, not a live feed. There's no webhook system and no batch endpoint for queuing hundreds of jobs at once. There's no service-level agreement either. If your workflow depends on any of those, it isn't there yet.
MCP, In One Sentence
MCP is a standard way for an AI assistant to use a tool directly, instead of telling you which buttons to press. With VocalLab's MCP server connected, an assistant can generate a voice line itself as part of finishing a task, rather than describing the steps and leaving you to carry them out.
A Concrete Example
Say you're working with an AI assistant on a video script. You finish a draft and ask the assistant to turn it into narration. With MCP connected, the assistant generates the audio itself and hands you back a finished file — you never open the app, cast a voice by hand, or copy the script anywhere. The assistant already has your saved voices and clones available to it, the same way the API does.
Same rules as the API: MCP is on every plan too, just like the API. It runs on the same rate limits and the same ⌈characters ÷ 15⌉ pricing underneath.
API or MCP — Which One Do You Need?
The API and MCP solve the same problem from two different directions, and picking between them is really about who's driving. Reach for the API when you're building your own app, script or pipeline and you want direct control over exactly when generation happens. Reach for MCP when you want an AI assistant to make that call itself, as one step inside a larger task it's already doing for you.
Plenty of teams end up using both: the API inside a product they're building, MCP for day-to-day work alongside an assistant.
- Building your own product or pipeline — Use the API — direct control, called from your own code.
- Working alongside an AI assistant — Use MCP — the assistant generates the line itself, as part of a task.
The Kind of Work This Unlocks
Content pipelines
A publishing tool that turns a finished article into narration automatically, without anyone opening VocalLab by hand.
Support automation
A support flow that reads a written reply back as audio for someone who'd rather listen than read.
Assistant-driven drafting
An AI assistant that finishes a script and hands back the narrated version in the same conversation.
Before You Connect
- Check your points balance — every call draws from it, whatever plan you're on
- Test with a short script first, since the ÷15 rate applies to every single request
- Know your rate limit going in: 60 requests a minute, 120 on Max
Where the Reference Docs Live
This post is deliberately light on implementation detail. Endpoint paths, parameter names, response formats and setup steps belong in a reference doc that gets kept current, not in a blog post that doesn't. Once you're signed in, the in-app API and MCP help articles walk through connecting either one step by step.
Who This Is For
Build voice generation directly into your own app or pipeline.
Connect MCP so an AI assistant can generate narration as part of a task, not just describe how to.
Reach your clones programmatically
Send scripts through the API and get them back in a voice you've already cloned.
Plan around the rate limits and per-character pricing before you queue a large run.
Hear the Voices Behind the API
Every voice reachable through the app is reachable through the API and MCP too. Here are six worth auditioning — the same catalog your own code or assistant will be casting from. Browse the full voice library for the rest.
Technical walkthroughs, product explainers
- Neutral American
- Clear, measured
Authoritative Male Tech Explainer Voice
Developer docs, tech reviews
- Neutral American
- Authoritative, confident
Articulate Male Explainer Voice
API walkthroughs, tutorials
- Neutral American
- Articulate, precise
Professional Clear American Male Voice
Support automation, status updates
- Neutral American
- Professional, clear
Warm Natural Female Explainer Voice
Product announcements, onboarding
- Neutral American
- Warm, natural
Curious Female Explainer Voice
Release notes, quick recaps
- Neutral American
- Curious, fast-paced
| Voice | Accent | Best for | Tone |
|---|---|---|---|
| Lucid Male Explainer Voice | Neutral American | Technical walkthroughs, product explainers | Clear, measured |
| Authoritative Male Tech Explainer | Neutral American | Developer docs, tech reviews | Authoritative, confident |
| Articulate Male Explainer Voice | Neutral American | API walkthroughs, tutorials | Articulate, precise |
| Professional Clear American Male | Neutral American | Support automation, status updates | Professional, clear |
| Warm Natural Female Explainer | Neutral American | Product announcements, onboarding | Warm, natural |
| Curious Female Explainer Voice | Neutral American | Release notes, quick recaps | Curious, fast-paced |
Frequently Asked Questions
What plan do I need for API or MCP access?
Any plan, Free included. Both spend points from your balance, so the plan you pick mostly decides how many points you get each month.
How is API usage priced?
By character count: characters divided by 15, rounded up, in points per generation. That's a deliberate premium over the app's divide-by-17 rate.
What are the rate limits?
60 requests per minute, rising to 120 requests per minute on Max.
Can the API reach my cloned or designed voices?
Yes. Both the API and MCP server reach the same catalog you use in the app, including your saved clones and designed voices.
Does the API support streaming or webhooks?
Not currently. A request returns a finished audio file. There's no streaming connection, webhook system, batch endpoint, or service-level agreement.