AI Voice Generator API

Overview

AI Voice Generator creates realistic speech audio from text using celebrity and character voices. The API produces high-quality voice synthesis with natural intonation, emotion, and speaking patterns that closely match famous personalities.

How It Works

Write the text - Enter the text you want spoken
Select a voice - Choose from available celebrity and character voices
API generates audio - AI synthesizes natural-sounding speech
Download the result - Retrieve your audio file

Use Cases

Content creation - Voice narration for videos and podcasts
Personalized messages - Custom voice messages and greetings
Entertainment - Create fun content with celebrity voices
Accessibility - Text-to-speech with engaging voices
Marketing - Voiceovers for ads and promotional content

Available Voices

Call List voices and pass the selected voice’s id as style.voice_name. Display names are not unique. Follow next_cursor with the cursor query parameter to fetch additional pages; limit defaults to 20 and accepts 1–100. Existing named voice values remain accepted. Supply text in style.prompt (up to 5,000 characters). Optional style.language accepts a canonical lowercase language name from the endpoint reference and defaults to auto. Optional style.speed accepts 0.75–1.30 and defaults to 1. See the API reference for the complete current language and voice options.

Best Practices

Text Guidelines

Write naturally - Use conversational language that sounds good when spoken aloud.

Natural punctuation - Use periods, commas, and question marks for natural pauses
Avoid abbreviations - Write out words fully (e.g., “Mister” not “Mr.”)
Short sentences - Break long text into shorter, digestible sentences
Read it aloud - Test how text sounds before generating

Optimizing Output Quality

Technique	Why It Helps
Natural punctuation	Creates appropriate pauses and intonation
Phonetic spelling	Helps with unusual words or names
Sentence breaks	Improves rhythm and comprehension
Consistent tone	Matches voice personality

Text Length Considerations

Short clips - 1-2 sentences work great for social media
Medium content - Paragraphs work well for narration
Long content - Consider breaking into multiple clips for variety

Code Examples

The Node.js examples create an audio project, then call audioProjects.checkResult to wait for completion and download the audio.

Basic Voice Generation

from magic_hour import Client
from os import getenv

client = Client(token=getenv("MAGIC_HOUR_API_KEY"))

result = client.v1.ai_voice_generator.generate(
    style={
        "prompt": "Hello! This is a test of the Magic Hour voice generator. Pretty cool, right?",
        "voice_name": "Morgan Freeman"
    },
    name="Voice Generator audio",
    wait_for_completion=True,
    download_outputs=True,
    download_directory="."
)

if result.status == "complete":
    print(f"✅ Voice generation complete!")
    print(f"Downloaded to: {result.downloaded_paths}")
    print(f"Credits charged: {result.credits_charged}")
else:
    print(f"❌ Job failed with status: {result.status}")
    if result.error:
        print(f"Error: {result.error.message}")

import { Client } from "magic-hour";

const client = new Client({ token: process.env.MAGIC_HOUR_API_KEY });

const project = await client.v1.aiVoiceGenerator.create({
  style: {
    prompt: "Hello! This is a test of the Magic Hour voice generator. Pretty cool, right?",
    voiceName: "Morgan Freeman",
  },
  name: "Voice Generator audio",
});

const result = await client.v1.audioProjects.checkResult(
  { id: project.id },
  {
    waitForCompletion: true,
    downloadOutputs: true,
    downloadDirectory: ".",
  }
);

console.log(`Status: ${result.status}`);
console.log(`Downloaded to: ${result.downloadedPaths}`);

Podcast-Style Voice

result = client.v1.ai_voice_generator.generate(
    style={
        "prompt": "Welcome to the show! Today we're going to talk about something really exciting.",
        "voice_name": "Joe Rogan"
    },
    name="Voice Generator audio",
    wait_for_completion=True,
    download_outputs=True,
    download_directory="."
)

if result.status == "complete":
    print(f"✅ Voice generation complete!")
    print(f"Downloaded to: {result.downloaded_paths}")
    print(f"Credits charged: {result.credits_charged}")
else:
    print(f"❌ Job failed with status: {result.status}")
    if result.error:
        print(f"Error: {result.error.message}")

const project = await client.v1.aiVoiceGenerator.create({
  style: {
    prompt: "Welcome to the show! Today we're going to talk about something really exciting.",
    voiceName: "Joe Rogan",
  },
  name: "Voice Generator audio",
});

const result = await client.v1.audioProjects.checkResult(
  { id: project.id },
  {
    waitForCompletion: true,
    downloadOutputs: true,
    downloadDirectory: ".",
  }
);

console.log(`Downloaded to: ${result.downloadedPaths}`);

Professional Narration

result = client.v1.ai_voice_generator.generate(
    style={
        "prompt": "In a world where technology meets creativity, Magic Hour brings your ideas to life.",
        "voice_name": "Donald Trump"
    },
    name="Voice Generator audio",
    wait_for_completion=True,
    download_outputs=True,
    download_directory="."
)

if result.status == "complete":
    print(f"✅ Voice generation complete!")
    print(f"Downloaded to: {result.downloaded_paths}")
    print(f"Credits charged: {result.credits_charged}")
else:
    print(f"❌ Job failed with status: {result.status}")
    if result.error:
        print(f"Error: {result.error.message}")

const project = await client.v1.aiVoiceGenerator.create({
  style: {
    prompt: "In a world where technology meets creativity, Magic Hour brings your ideas to life.",
    voiceName: "Donald Trump",
  },
  name: "Voice Generator audio",
});

const result = await client.v1.audioProjects.checkResult(
  { id: project.id },
  {
    waitForCompletion: true,
    downloadOutputs: true,
    downloadDirectory: ".",
  }
);

console.log(`Downloaded to: ${result.downloadedPaths}`);

Combining with Lip Sync

Generate a voice, then sync it to a video:

Python

from magic_hour import Client
from os import getenv

client = Client(token=getenv("MAGIC_HOUR_API_KEY"))

# Step 1: Generate the voice
voice_result = client.v1.ai_voice_generator.generate(
    style={
        "prompt": "This is my custom voiceover for the video.",
        "voice_name": "Elon Musk",
    },
    name="Voice Generator audio",
    wait_for_completion=True,
    download_outputs=True,
    download_directory=".",
)

if voice_result.status != "complete" or not voice_result.downloaded_paths:
    raise RuntimeError(f"Voice generation did not complete: {voice_result.status}")

# Step 2: Use the generated audio for lip sync
lip_sync_result = client.v1.lip_sync.generate(
    assets={
        "video_file_path": "https://raw.githubusercontent.com/runshouse/Sample_Assets/main/sideeyegirl.mp4",
        "audio_file_path": voice_result.downloaded_paths[0],  # local path now exists
        "video_source": "file",
    },
    start_seconds=0,
    end_seconds=2,
    max_fps_limit=30,
    style={"generation_mode": "lite"},
    name="Lip Synced Video",
    wait_for_completion=True,
    download_outputs=True,
    download_directory=".",
)

print("voice_status:", voice_result.status, "paths:", getattr(voice_result, "downloaded_paths", None))
print("lip_status:", lip_sync_result.status, "paths:", getattr(lip_sync_result, "downloaded_paths", None))

API pricing

Voice generation costs 0.1 credits per character of text, rounded up to the nearest whole number. Inputs support up to 5,000 characters.

Text Length	Credits
100 characters (1-2 sentences)	10 credits
500 characters (a paragraph)	50 credits
5,000 characters (maximum)	500 credits

Try this in our Google Colab Cookbook: Run this API with sample code. Just add your API key.

API Reference