Deepgram Real-time Transcription

A real-time speech-to-text application using Deepgram's live transcription API with WebSocket support.

Features

• Real-time speech transcription
• WebSocket communication for live audio streaming
• Support for multiple languages and models
• Smart formatting and punctuation
• Speaker diarization
• Custom keyword detection for specific terms

Prerequisites

• Node.js (v14 or higher)
• Deepgram API key

Installation

1. Clone the repository:

git clone <your-repo-url>
cd deepgram-test

2. Install dependencies:

npm install

3. Create a .env file in the root directory and add your Deepgram API key:

DEEPGRAM_API_KEY=your_deepgram_api_key_here

Usage

1. Start the server:

node index.js

2. Open your browser and navigate to http://localhost:3000

3. Click the "Start Recording" button to begin real-time transcription

Configuration

The application supports various Deepgram features:

• Model: Choose from different Deepgram models
• Language: Supports multiple languages (default: en-US)
• Smart Formatting: Automatically formats numbers, dates, and punctuation
• Diarization: Identifies different speakers
• Custom Keywords: Detects specific terms like "Pinn", "Dr. Pinnamaneni", etc.

API Endpoints

• GET / - Serves the main application page
• WebSocket / - Handles real-time audio streaming and transcription

WebSocket Messages

Client to Server:

• { type: "start", model: "nova-2", language: "en-US", smart_format: true } - Start transcription
• { type: "audio", audio: "base64_encoded_audio" } - Send audio data
• { type: "stop" } - Stop transcription

Server to Client:

• Transcription results from Deepgram API

Environment Variables

• DEEPGRAM_API_KEY - Your Deepgram API key (required)

Dependencies

• @deepgram/sdk - Deepgram SDK for Node.js
• express - Web framework
• ws - WebSocket library
• dotenv - Environment variable management

License

ISC

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request