Burki - The Open-Source Voice AI Platform

Build and deploy production-ready, multi-tenant AI voice assistants in minutes, not months.

Burki is the open-source alternative to vapi.ai that actually delivers on its promises. Unlike proprietary platforms that nickel-and-dime you with complex pricing and poor performance, Burki provides a complete, production-ready voice AI platform with ultra-low latency (0.8-1.2s vs vapi.ai's 4-5s), a beautiful web interface that works, and transparent costs through self-hosting.

Experience the difference yourself: See Burki's sub-second latency and beautiful interface in action, then compare it to vapi.ai's sluggish performance. The difference is immediately obvious.

The Open-Source Alternative to vapi.ai That Actually Works

Tired of vapi.ai's complex pricing, poor latency, and developer-only complexity? Burki delivers what voice AI platforms should have been from the start.

• Ultra-Low Latency: 0.8-1.2 seconds vs vapi.ai's 4-5+ seconds
• Crystal Clear Audio: Built-in RNNoise for real-time audio denoising
• Production-Ready: Multi-tenant architecture that scales to real-world call volumes

• No Hidden Costs: Open-source means no surprise billing or complex pricing tiers
• All-in-One: Web dashboard, analytics, and management tools included out of the box
• Self-Hosted: Complete control over your costs and data

• Beautiful UI: A web interface that actually works (unlike vapi.ai's notorious UI issues)
• Non-Technical Friendly: Manage assistants without deep developer expertise
• Complete Platform: Everything you need in one place, not scattered across multiple services

Real User Experience: "I was a vapi.ai power user for 2.5 years. The UI never worked properly, latency was terrible (4-5 seconds), and their recent pricing changes would cost my startup $800/month. With Burki, I get sub-second response times and complete control over my costs."

• End-to-End Call Handling: Full lifecycle management from incoming call to detailed post-call analysis
• Multi-Assistant Support: Create unlimited assistants with unique personalities, voices, and specialized knowledge
• Real-time Conversations: WebSocket streaming with 0.8-1.2 second response times (5x faster than vapi.ai)
• Crystal Clear Audio: Built-in RNNoise denoising that actually works out of the box

• Multi-Tenant Design: Support unlimited organizations with complete data isolation
• Scalable Infrastructure: Handle thousands of concurrent calls with auto-scaling
• Beautiful Web Dashboard: Manage everything through an interface that actually works
• Advanced Analytics: Real-time monitoring, call success rates, and detailed performance metrics

• Telephony: Twilio with WebSocket streaming
• LLM Providers: OpenAI, Anthropic, Gemini, xAI, Groq, and custom providers
• TTS Providers: ElevenLabs, Deepgram, Inworld, Resemble, OpenAI
• STT Providers: Deepgram Nova with confidence scoring
• Knowledge Base (RAG): Upload documents to make assistants smarter

• Real-time Noise Reduction: RNNoise integration for broadcast-quality calls
• Voice Activity Detection: Smart silence detection for natural conversations
• Call Recording: Automatic recording with transcript storage and search
• Background Sound Support: Add ambiance for realistic call environments

• RESTful API: Complete programmatic control over all platform features
• Webhook Support: Real-time notifications for call events and integrations
• No-Code Assistant Creation: Build sophisticated voice agents without programming
• Custom Tool Integration: Connect to external APIs and databases
• Secure Authentication: OAuth, API keys, and role-based access control

Get a full Burki instance running locally with a single command.

Prerequisites: Docker & Docker Compose

1. Clone the repository:

git clone https://github.com/meeran03/burki.git
cd burki

2. Configure your environment:

cp .env.example .env

Now, open the `.env` file and add your API keys for Twilio, your chosen LLM, TTS, and STT providers.

3. Deploy!

chmod +x deploy.sh
./deploy.sh

This script will build the Docker images and start the application and database using Docker Compose.

🎉 That's it! Your Burki instance is now running.

• Web Dashboard: http://localhost:8000
• API Docs: http://localhost:8000/docs

If you prefer to run the application directly on your host machine:

Prerequisites: Python 3.11+, PostgreSQL

1. Clone and install dependencies:

    git clone https://github.com/meeran03/burki.git
    cd burki
pip install -r requirements.txt

2. Configure your environment:

   cp .env.example .env
   # Edit the .env file with your credentials and database URL

3. Set up the database: This command runs all necessary database migrations.

   alembic upgrade head

4. Run the application:

   # For development
   uvicorn app.main:app --reload
   
   # For production
   gunicorn app.main:app --bind 0.0.0.0:8000 --worker-class uvicorn.workers.UvicornWorker

Contributions are welcome! Please feel free to fork the repository, make changes, and submit a pull request. For major changes, please open an issue first to discuss what you would like to change.

This project is licensed under the MIT License. See the LICENSE file for details.

Meeran Malik

• Portfolio: meeran.dev
• Twitter/X: @evolvinginsaan
• LinkedIn: Meeran Malik
• GitHub: @meeran03