Skip to content

Latest commit

 

History

History
122 lines (84 loc) · 3.66 KB

README.md

File metadata and controls

122 lines (84 loc) · 3.66 KB

Pipecat Starter Kit: Real-time Voice AI with Pipecat

Welcome to the Pipecat Starter Kit! This repository is designed to help you quickly get started with the Pipecat Framework, providing a comprehensive boilerplate for building real-time voice AI applications. Whether you're new to Pipecat or looking to streamline your development process, this starter kit has you covered.

Overview

Pipecat can be complex for newcomers, but this starter kit simplifies the process by integrating essential features and services into a single package. Here's what you'll find:

  • Sentry for Metrics Logging: Monitor and log application metrics seamlessly.
  • Transcription Streaming: Real-time transcription streaming to the frontend.
  • Back-Channeling: Enhance conversational flow with back-channeling capabilities.
  • Function Calling: Execute functions dynamically within the conversation.
  • Multilingual Support: Engage users in multiple languages.
  • Vapi Compoenents: Extra rizz

The backend leverages Pipecat Flows for conversation management, while the frontend is built using the Pipecat React SDK.

Quick Start

Follow these steps to set up your environment and start developing:

Backend Setup

cd bots
python -m venv venv
source venv/bin/activate  # or `venv\Scripts\activate` on Windows
pip install -r requirements.txt
cp .env.example .env  # Edit with your API keys

# Start the server using UV
uv run python server.py

Frontend Setup

cd react
npm install
npm run dev

Architecture

Backend

The backend is structured around a modular pipeline architecture, ensuring minimal latency and efficient processing:

pipeline = [
    transport.input(),      # WebRTC input stream
    rtvi_speaking,         # Voice activity detection
    stt,                  # Speech-to-text conversion
    rtvi_user_transcript, # Real-time user transcript processing
    context_aggregator,   # Message context management
    llm,                 # Language model processing
    rtvi_bot_llm,       # Real-time LLM output processing
    tts,               # Text-to-speech conversion
    rtvi_bot_tts,    # Real-time TTS word streaming
    transport.output() # WebRTC output stream
]

Frontend

The React frontend uses a layered approach for real-time audio visualization and transcription:

<RTVIProvider>              // WebRTC connection management
  <BotTypeProvider>         // Bot selection context
    <RadialCard />          // Audio visualization
    <Transcriber />         // Real-time transcription
    <RTVIClientAudio />     // Audio I/O management
  </BotTypeProvider>
</RTVIProvider>

Environment Setup

Ensure you have the following environment variables configured:

# API Keys
OPENAI_API_KEY=your_key_here
DAILY_API_KEY=your_key_here
DEEPGRAM_API_KEY=your_key_here
GLADIA_API_KEY=your_key_here

# Configuration
DAILY_ROOM_URL=your_room_url  # Optional, will create if not provided
FAST_API_PORT=7860

Additional Environment Variables

Depending on the bot type you're using, you may need to configure additional variables:

Debugging Tools

  • Pipeline Debugging: Log pipeline activities for troubleshooting.
  • Sentry Statistics: Monitor TTFB for the different services you're using

Contributing

We welcome contributions! Please follow these steps:

  1. Fork the repository
  2. Create a feature branch
  3. Add tests for new features
  4. Submit a pull request

License

This project is licensed under the BSD 2-Clause License.

For more detailed information about specific components or advanced usage patterns, please refer to the official Pipecat documentation.