VoiceMix

VoiceMix is a flexible text-to-speech library that allows you to generate speech from text using different voices, languages, and customization options.

Installation

npm install voicemix dotenv

Environment Setup

Create a .env file in your project root with your API keys:

ELEVENLABS_API_KEY="6e04xxxxxxxxxxxxxxxxxxxxxxxxa9da"
RESEMBLE_API_KEY="9YWxxxxxxxxxxxxxxxxxmgtt"
CARTESIA_API_KEY="sk_car_xxxxxxxxxxxxxxxxxxxxjr"

Usage

Basic Example

import { VoiceMix } from 'voicemix';
import dotenv from 'dotenv';
dotenv.config();

const voiceMix = new VoiceMix();

voiceMix
    .voice('EbhcCfMvNsbvjN6OhjpJ')
    .say('Hello, world!')
    .save();

Using ElevenLabs v3 Model

const voiceMix = new VoiceMix();

voiceMix
    .v3() // Use the latest ElevenLabs v3 model
    .voice('EbhcCfMvNsbvjN6OhjpJ')
    .say('Hello! This is using the ElevenLabs v3 model.')
    .save();

The v3 model is the latest and most advanced model from ElevenLabs, providing the most natural and expressive voice generation.

Advanced Usage

Using Resemble AI

const voiceMix = new VoiceMix();

voiceMix
    .useResemble()  // https://www.resemble.ai/
    .prompt('Friendly and conversational tone') // Set voice prompt/style
    .voice('ba875a0a') // Select specific voice
    .lang('en-US') // Set language
    .say('Your text here')
    .save();

Using Cartesia

const voiceMix = new VoiceMix();

voiceMix
    .useCartesia()  // https://cartesia.ai/
    .voice('your-cartesia-voice-id') // Select specific voice from your Cartesia account
    .say('Your text here')
    .save();

Note: You need to use a valid voice ID from your Cartesia account. You can find available voices in your Cartesia dashboard.

Batch Processing

You can process multiple lines of text using a JSON configuration:

import { VoiceMix } from 'voicemix';
import fs from 'fs';

// Read script from JSON file
const script = JSON.parse(fs.readFileSync('./lines.json', 'utf8'));
const voiceMix = new VoiceMix();

// Process each line
for (const entry of script) {
    voiceMix
        .prompt(entry.prompt || 'Friendly and conversational tone')
        .voice(entry.voiceId)
        .say(entry.english)
        .save();
}

Example lines.json:

[
    {
        "prompt": "Friendly and conversational tone",
        "english": "Hello, how are you today?",
        "voiceId": "EbhcCfMvNsbvjN6OhjpJ"
    }
]

Features

• Multiple voice support
• Language selection
• Voice prompts for style control (Resemble AI)
• Support for multiple TTS providers:
  • ElevenLabs (including v3 model)
  • Resemble AI
  • Cartesia
• Support for different ElevenLabs models:
  • monolingual_v1() - Original English model
  • multilingual_v1() - First multilingual model
  • multilingual_v2() - Improved multilingual model (default)
  • v3() - Latest and most advanced model
• Simple chainable API

License

MIT