Give any application a voice.

The JSF Labs API exposes two capabilities behind one gateway. You send audio and get back a reusable voice; you send text and get back speech in that voice. Both endpoints share the same authentication, rate limits, character-based billing, and error format, so once you've called one you already know how to call the other.

• Clone a voice — upload a short sample and receive a portable jsf_ voice ID.
• Text to speech — pass that voice ID with your text and receive a hosted audio file, loudness-normalized and peak-safe.

Everything is synchronous: each request returns the finished result, so there are no jobs to poll and no webhooks to configure.

All endpoints live under a single versioned base. Every path in these docs is written in full, including the version prefix.

https://www.jsflabs.io/api/v1

Requests must be sent over HTTPS. Plain HTTP is not supported.

Clone a voice, then speak with it — two calls, start to finish.

1 — Clone a voice from a sample

curl -X POST https://www.jsflabs.io/api/v1/voice-clone \
  -H "x-api-key: jsf_your_api_key" \
  -F "audio=@sample.wav" \
  -F "name=Morgan" \
  -F "gender=Female" \
  -F "language=en"

2 — Speak with the cloned voice

curl -X POST https://www.jsflabs.io/api/v1/text-to-speech \
  -H "x-api-key: jsf_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "The quick brown fox jumps over the lazy dog.",
    "voice_id": "jsf_c3BlZWNoaWZ5X3ZvaWNlX2lk",
    "speed": 1
  }'

Every request is authenticated with your secret API key, sent in the x-api-key header. Keys always begin with jsf_. You can find and manage keys in your JSF Labs dashboard.

x-api-key: jsf_your_api_key

Requests with a missing key, a malformed key, an unknown key, an inactive subscription, or an expired key are rejected before any work is done. See Error codes for the exact responses.

Each API key may make up to 50 requests per minute, measured as a sliding window. The limit is per key, so separate keys have separate budgets.

When you exceed it, the API responds with 429 and tells you how long to wait:

Retry-After: 37
X-RateLimit-Limit: 50
X-RateLimit-Remaining: 0

{
  "success": false,
  "error": {
    "code": "RATE_LIMITED",
    "message": "Rate limit exceeded. You can make up to 50 requests per minute.",
    "retryAfterSeconds": 37
  }
}

Read the Retry-After header and back off for that many seconds before retrying. Need a higher ceiling? Contact support@jsflabs.io.

Usage is metered in characters. Your key carries a total allocation and a running used count; the difference is your available balance.

• Text to speech costs one character per character of text in the request.
• Cloning is currently free — it costs zero characters.

Every successful response echoes your balance so you can track it without a separate call:

"credits": {
  "cost": 137,        // charged by this request
  "used": 137,        // total used so far
  "remaining": 49863,  // what's left
  "total": 50000       // your allocation
}

If a request would exceed your balance, it's rejected with 402 and INSUFFICIENT_CHARACTERS before any audio is generated — you're never charged for a request you couldn't afford.

Cloning returns a portable voice identifier prefixed with jsf_. This is the only identifier you ever need — pass it straight into text to speech as voice_id.

jsf_c3BlZWNoaWZ5X3ZvaWNlX2lk

Treat the value as opaque — don't parse or modify it. A value that doesn't start with jsf_, or that has been altered, is rejected with INVALID_VOICE_ID.

Upload a clean audio sample and receive a reusable voice. Send the request as multipart/form-data.

Body — multipart/form-data

Request

curl -X POST https://www.jsflabs.io/api/v1/voice-clone \
  -H "x-api-key: jsf_your_api_key" \
  -F "audio=@sample.wav" \
  -F "name=Morgan" \
  -F "gender=Female" \
  -F "language=en"

Response · 201 Created

{
  "success": true,
  "message": "Your voice has been cloned successfully.",
  "data": {
    "voiceId": "jsf_c3BlZWNoaWZ5X3ZvaWNlX2lk",
    "name": "Morgan",
    "language": "en",
    "gender": "Female",
    "provider": "speechify",
    "sampleUrl": "https://cdn.jsflabs.io/clones/.../sample.wav",
    "createdAt": "2026-08-12T09:20:44.120Z"
  },
  "credits": { "cost": 0, "used": 0, "remaining": 50000, "total": 50000 }
}

Turn text into speech in a cloned voice. Send the request as application/json. The response is a hosted, loudness-normalized WAV file.

Body — application/json

Request

curl -X POST https://www.jsflabs.io/api/v1/text-to-speech \
  -H "x-api-key: jsf_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "The quick brown fox jumps over the lazy dog.",
    "voice_id": "jsf_c3BlZWNoaWZ5X3ZvaWNlX2lk",
    "speed": 1
  }'

Response · 200 OK

{
  "success": true,
  "message": "Your audio has been generated successfully.",
  "data": {
    "audio_url": "https://cdn.jsflabs.io/tts/.../a1b2c3.wav",
    "voice_id": "jsf_c3BlZWNoaWZ5X3ZvaWNlX2lk",
    "format": "wav",
    "characters": 44,
    "speed": 1
  },
  "credits": { "cost": 44, "used": 44, "remaining": 49956, "total": 50000 }
}

Every response uses the same envelope, so you can branch on one field. Success and failure never share a shape.

Success

{ "success": true, "message": "...", "data": { ... }, "credits": { ... } }

Failure

{
  "success": false,
  "error": {
    "code": "VALIDATION_ERROR",
    "message": "Some fields are invalid. Check the fields object for details.",
    "fields": { "audio": "Unsupported audio format. Please upload a WAV or MP3 file." }
  }
}

Always branch on success. On failure, error.code is a stable machine-readable string you can switch on; error.message is human-readable and safe to surface to your users. Some errors add extra fields (like fields or retryAfterSeconds).

The full set of error codes, grouped by what caused them. Codes are stable — safe to match on in your integration.

Authentication & access

Request & quota

Cloning

Text to speech

Server