Text to Speech Guide

Learn how to convert text to speech audio with LEAPERone, choose TTS models and voices, and handle binary audio responses.

The Text to Speech endpoint generates spoken audio from text input. Choose from preset voices across multiple languages, or clone any voice from a short audio sample.

Models

Model	Pricing	Best for
`moss-tts-nano`	0.005 credits / 1K chars	Multilingual speech with voice cloning

Quick Start

POST /v1/audio/speech

curl -X POST https://api.leaper.one/v1/audio/speech \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"input": "Hello world!", "voice": "alloy"}' \
  --output hello.wav

The response is a WAV audio file you can play directly.

Using the OpenAI SDK

The endpoint is OpenAI-compatible, so you can use the official SDK:

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.leaper.one/v1",
    api_key="sk-your-api-key"
)

response = client.audio.speech.create(
    model="moss-tts-nano",
    input="Welcome to LEAPERone. This is a text to speech demo.",
    voice="alloy"
)
response.stream_to_file("welcome.wav")

Node.js

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.leaper.one/v1",
  apiKey: "sk-your-api-key",
});

const response = await client.audio.speech.create({
  model: "moss-tts-nano",
  input: "Welcome to LEAPERone.",
  voice: "alloy",
});

const buffer = Buffer.from(await response.arrayBuffer());
await fs.promises.writeFile("welcome.wav", buffer);

Choosing a Voice

OpenAI-compatible voices

Use familiar voice names if you're migrating from OpenAI:

Voice	Maps to
`alloy`	English, neutral
`echo`	English, news anchor
`fable`	English, gentle
`onyx`	English, academic
`nova`	Chinese, neutral
`shimmer`	Chinese, soft

Language-specific voices

Voices are named by language and style, e.g. zh-gentle, en-news, ja-news.

Chinese gentle voice

curl -X POST https://api.leaper.one/v1/audio/speech \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"input": "晚安，祝你做个好梦。", "voice": "zh-gentle"}' \
  --output goodnight.wav

Japanese news voice

curl -X POST https://api.leaper.one/v1/audio/speech \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"input": "本日のニュースをお伝えします。", "voice": "ja-news"}' \
  --output news_ja.wav

See the API Reference for the full list of available voices.

Voice Cloning

Upload a short audio sample to clone any voice. This requires multipart/form-data instead of JSON.

Clone a voice

curl -X POST https://api.leaper.one/v1/audio/speech \
  -H "Authorization: Bearer sk-your-api-key" \
  -F input="This sentence will be spoken in the cloned voice." \
  -F prompt_audio=@my-voice-sample.mp3 \
  --output cloned.wav

Tips for good voice cloning:

Use a clear speech recording, 5-15 seconds long
Minimize background noise
Audio file must be under 1MB (mp3, wav, flac, m4a, ogg)
Voice cloning runs on CPU and takes 20-60 seconds

Voice cloning is not supported via the OpenAI SDK because it requires file upload alongside text parameters. Use curl or a direct HTTP request.

Output Formats

Format	Use case
`wav` (default)	Ready-to-play audio file
`pcm`	Raw audio for further processing (48kHz, 16-bit, stereo)

Get PCM output

curl -X POST https://api.leaper.one/v1/audio/speech \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"input": "Raw PCM output.", "voice": "alloy", "response_format": "pcm"}' \
  --output raw.pcm

Billing is based on character count. See the API Reference for pricing details.

On this page