voiceflow-gallagan / elevenlabs-integration

VF-ElevenLabs API Integration

This Node.js application uses the Eleven Labs API to synthesize text to speech. It takes in text and voice settings as input and returns the synthesized audio in a data URI format you can use in an Audio step in your Assistant.

Loom video: https://www.loom.com/share/a9a0abf118594ba590386024c905596f

Prerequisites

• Node.js installed on your machine

Installation

1. Clone the repository:

   git clone https://github.com/voiceflow-gallagan/VF-ElevenLabs.git
   

2. Change to the project directory:

   cd VF-ElevenLabs
   

3. Install the required dependencies:

   npm install
   

4. Replace your_api_key_here in the .env.template file with your Eleven Labs API key.

   ELEVENLABS_API_KEY=your_api_key_here
   PORT=3000
   

5. Rename the .env.template file to .env.

Running the App

Run the following command in the project directory:

npm start

The server will start listening on the specified port (default is 3000). You can now make a POST request to the /synthesize endpoint with the text and voice settings as input.

API Endpoint

POST /synthesize

Request Body:

• text(required): The text to synthesize.
• model (optional): The model you want to use. Default is 'eleven_multilingual_v2'
• voice (optional): The voice to use for synthesis. Default is '21m00Tcm4TlvDq8ikWAM'.
• voice_settings (optional): An object containing additional voice settings.
  • stability (default: 0)
  • similarity_boost (default: 0)

Response:

Returns an object containing the synthesized audio in a data URI format.

{
  "audioDataURI": "data:audio/mpeg;base64,..."
}

Example Request

curl -X POST -H "Content-Type: application/json" \
  -d '{"text": "Hello, world!", "voice": "21m00Tcm4TlvDq8ikWAM", "voice_settings": {"stability": 0, "similarity_boost": 0}}' \
  http://localhost:3000/synthesize

Example Response

{
  "audioDataURI": "data:audio/mpeg;base64,//uQZAAAAAAAAAAAAAAAAAAAAA"
}