Text to Speech API with a Director's Note

Three things other speech APIs make you work for

A director's note, in plain English

Pass "note": "warm, unhurried" with the text and the voice follows it: emotion, pace, accent, where to breathe. No SSML, no per-sentence tuning. Works on the Ultra and Pro tiers, and emotion tags like [whispers] work inline.

OpenAI-compatible

Already calling /v1/audio/speech? Change base_url to https://api.leapfun.ai/v1, keep your SDK, and the instructions field becomes the director's note. Model names map to our tiers.

One key, shared credits

API calls spend the same credits as the app, at the same rates, with no separate API price list. Create a key in Settings, read your balance at GET /v1/usage. Every plan includes it, Free too; higher plans get higher rate limits. Commercial use of the audio starts on the Starter plan.

Quickstart

One request, one MP3. Replace lf_sk_... with a key from Settings → Developer API in the app.

terminal

curl https://api.leapfun.ai/v1/tts \
  -H "Authorization: Bearer lf_sk_..." \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Welcome back. We pick up right where we left off.",
    "engine": "ultra",
    "voice": "aurora",
    "note": "warm, unhurried"
  }' \
  --output welcome.mp3

The body is MP3 bytes. The headers tell you what it cost: X-Job-Id, X-Credits-Used, X-Engine, and X-RateLimit-Limit / -Remaining / -Reset.

speak.py · pip install openai

from openai import OpenAI

client = OpenAI(base_url="https://api.leapfun.ai/v1", api_key="lf_sk_...")

speech = client.audio.speech.create(
    model="gpt-4o-mini-tts",   # runs on Leap Ultra
    voice="aurora",
    input="Welcome back. We pick up right where we left off.",
    instructions="warm, unhurried",   # the director's note
)
speech.write_to_file("welcome.mp3")

The official openai package, unchanged. Only base_url and the key are ours.

speak.mjs · npm install openai

import OpenAI from "openai";
import { writeFile } from "node:fs/promises";

const client = new OpenAI({ baseURL: "https://api.leapfun.ai/v1", apiKey: "lf_sk_..." });

const speech = await client.audio.speech.create({
  model: "gpt-4o-mini-tts",   // runs on Leap Ultra
  voice: "aurora",
  input: "Welcome back. We pick up right where we left off.",
  instructions: "warm, unhurried",   // the director's note
});
await writeFile("welcome.mp3", Buffer.from(await speech.arrayBuffer()));

Works the same from Deno, Bun or an edge function: it is one HTTPS request.

Install the nodeSettings → Community Nodes → install n8n-nodes-leapfun.
Add the credentialPaste your key from Settings → Developer API in the app. Leave Base URL at its default.
Speech → GenerateEngine Ultra, voice aurora, Director's Note warm, unhurried. The MP3 lands in the binary field data.

No community node allowed on your instance? An HTTP Request node does the same: POST https://api.leapfun.ai/v1/tts, header Authorization: Bearer lf_sk_..., JSON body with text, engine, voice, note, response format File. The full guide is on LeapFun for n8n.

Endpoints

Base URL https://api.leapfun.ai/v1. Every request carries Authorization: Bearer lf_sk_.... Up to 5,000 characters of text per request.

Endpoint	What it does	Returns
POST /v1/tts	Generate speech. Body: `text`, `engine` (`ultra` · `pro` · `expressive` · `turbo`, default `ultra`), `voice`, optional `note`.	MP3 bytes + `X-Job-Id`, `X-Credits-Used`, `X-Engine`, `X-RateLimit-*`
POST /v1/audio/speech	OpenAI-compatible speech. Body: `model`, `voice`, `input`, optional `instructions` (the director's note).	MP3 bytes, same headers
GET /v1/voices	The voice catalog with ids, languages and the engines each voice runs on. Your cloned and designed voices are included.	JSON
GET /v1/engines	The four engines, what each bills and which one your plan can use.	JSON
GET /v1/usage	Your credit balance and this period's spend, app and API together.	JSON

Model names on the OpenAI-compatible route

You send	Runs on	Also accepted
`gpt-4o-mini-tts`	Leap Ultra	`leap-ultra`
`tts-1-hd`	Leap Pro	`leap-pro`
—	Leap Expressive	`leap-expressive`
`tts-1`	Leap Turbo	`leap-turbo`

Rate limits and billing

Limits are per workspace and per minute. Each response reports them in X-RateLimit-Limit, X-RateLimit-Remaining and X-RateLimit-Reset; a 429 carries Retry-After.

Plan	Requests per minute	API access
Free	10	Included, personal use
Starter	20	Included
Creator	30	Included
Pro	120	Included
Team	300	Included, shared by the team's seats

What a request costs

The same credits and the same rates as the app. There is no API surcharge, and a key does not change your plan. Plans and credit allowances are on the pricing page.

Engine	Bills	Director's note
Leap Ultra	1.5 credits per UTF-8 byte	Yes
Leap Pro	1 credit per character; voices shared with Ultra bill 1 credit per UTF-8 byte	Yes
Leap Expressive	1 credit per character	Emotion tags only
Leap Turbo	0.5 credits per character	No

Characters are counted on the text you send, including spaces and emotion tags. X-Credits-Used on every response is the exact figure.

Errors

Every error is JSON with one error object: type, code, message, param (the field at fault, when there is one) and request_id to quote to support.

Status	`error.code`	Meaning
401	`invalid_api_key`	The key is missing, revoked or mistyped.
402	`insufficient_credits`	Not enough credits for this request. Top up or wait for the monthly refill.
403	`plan_required`	Your workspace role cannot generate audio, or API access is switched off for your plan. Ask a workspace admin.
413	`text_too_long`	More than 5,000 characters in one request. Split the text.
429	`rate_limited`	Over your plan's requests per minute. Wait for `Retry-After` seconds.

402 response

{
  "error": {
    "type": "billing_error",
    "code": "insufficient_credits",
    "message": "This request needs 1,284 credits; the workspace has 310.",
    "param": null,
    "request_id": "req_01J9W2K7H4"
  }
}

Questions developers ask

Which plan includes the API?

All of them, including Free. Plans differ in credits and in requests per minute (Free 10, Starter 20, Creator 30, Pro 120, Team 300). Audio made on the Free plan is for personal, non-commercial use; commercial rights start on Starter.

Does the API cost extra?

No. There is no API price list. A request spends the same credits the same text would cost in the editor, from the same balance, and GET /v1/usage shows both together.

Is it OpenAI-compatible?

For speech, yes. Point the official SDK at https://api.leapfun.ai/v1 and call audio.speech.create as before; gpt-4o-mini-tts, tts-1-hd and tts-1 map to Ultra, Pro and Turbo, and instructions is read as the director's note. Our own /v1/tts exposes the extra response headers.

Can I use my cloned voice?

Yes. Voices you cloned or designed in the app appear in GET /v1/voices with their ids and run on the engines they were made for. The usual plan rules for cloning apply; the API adds none of its own.

Tell us what you're building

We are shaping the API with the people who use it. Leave an email and a line about your project and we will write back when there is something to show, including early access to new endpoints and higher limits.

Ready to ship a voice?

Create a key in Settings, make the first request above, and hear your note followed.

Create an API key