API Overview

The VocalLab API lets you generate speech, list voices, and check your balance programmatically — from your own apps, scripts, or automation tools.

Watch the full walkthrough — covers the API and MCP

Available on every plan, Free included. Each call spends your account's points (1 point per 15 characters). Create a key in Studio under API & MCP in the sidebar (or on the Developer page on vocallab.ai). (AppSumo lifetime licences: Tier 3 and above.)

Base URL

https://api.vocallab.ai

Authentication

Send your API key as a Bearer token on every request:

Authorization: Bearer vl_live_...

Keys are created and revoked in Studio under API & MCP (or on the Developer page). Treat a key like a password — it's shown once, and anyone with it can spend your points. See API Keys.

What you can do

• Generate speech — POST /api/v1/tts (optionally pass a model)
• List voices — GET /api/v1/voices (catalog + your cloned & designed voices)
• List models — GET /api/v1/models (VocalLab Studio / Studio Flash / Pro / Lite)
• Check status / get the hosted URL — GET /api/v1/tts/:id
• Download SRT captions — GET /api/v1/tts/:id/captions (subtitle lines or word-by-word)
• Design a voice — POST /api/v1/voices/designs/previews, then POST /api/v1/voices/designs to keep one
• Clone a voice — POST /api/v1/voices/clones (one short recording, sent as base64)
• Check your balance — GET /api/v1/ping or GET /api/v1/me

Choosing a model

Pass "model" on POST /api/v1/tts to pick the engine (default v-pro):

• v-studio — VocalLab Studio: newest, steerable (follows expression instructions), 200+ languages.
• v-flash — VocalLab Studio Flash: same 200+ languages, about 5× faster, bills at ¾ the points. Not steerable.
• v-pro — VocalLab Pro: highest fidelity, 15 languages. The default.
• v-lite — VocalLab Lite: fast and economical — bills at half the points.

Your API key can use all four. See Voice Models for details.

See the API Reference for full details.

Billing

API generations spend from the same points balance as the web app. The cost is calculated from your text length, not the final audio duration:

• API: points = ⌈ characters ÷ 15 ⌉ — about 15 characters per point.
• The web app bills at 17 characters per point, so the API carries a small (~13%) premium for programmatic access.

points_used is returned on every generation. Points are metered before the voice is generated, so you can never spend more than your balance — out-of-points requests return 402. Since spoken audio averages ~15–17 characters/second, 1 point ≈ 1 second of audio is a close rule of thumb, but the exact charge always follows the character formula above.

Use it inside AI tools

Prefer Claude, Cursor, or another AI assistant? The MCP server exposes the same capabilities as native tools — nothing to install with the hosted option.