Voice Pizza Ordering Chatbot

A Python voice chatbot that acts as "Bob the pizza guy" and handles pizza-order conversations end-to-end using speech input and spoken responses.

It combines:

• Speech-to-text with Whisper
• Chat completion with Groq (llama3-8b-8192)
• Text-to-speech with OpenAI (tts-1)
• Audio playback with pygame

Demo

• Video demo: Chatbot Demo

Features

• Voice-based user input from microphone
• Persona-driven pizza ordering assistant prompt
• Conversation memory across turns in the same session
• Spoken assistant responses via generated assistant.mp3

Project Structure

myChatbot/
├── README.md
└── chatbot/
    ├── main.py
    ├── requirements.txt
    └── .gitignore

Tech Stack

• Python 3.9+
• groq SDK for LLM chat responses
• openai SDK for text-to-speech
• SpeechRecognition + whisper + soundfile + numpy for transcription
• pygame for local audio playback
• python-dotenv for environment variables

Prerequisites

• Python 3.9 or newer
• Working microphone
• API keys:
  • GROQ_API_KEY
  • OPENAI_API_KEY

Setup

1. Clone the repository and go to the project:

   git clone <your-repo-url>
   cd myChatbot/chatbot

2. Create and activate a virtual environment:

   python -m venv .venv
   source .venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

4. Create a .env file in chatbot/:

   GROQ_API_KEY=your_groq_api_key_here
   OPENAI_API_KEY=your_openai_api_key_here

Run the Chatbot

From the chatbot/ directory:

python main.py

Then speak into your microphone when prompted with say something....

How It Works

1. Captures audio from the microphone.
2. Converts speech to text with Whisper.
3. Sends the transcribed text to Groq chat completion.
4. Converts assistant response text to speech using OpenAI TTS.
5. Plays the generated response audio locally with pygame.

Notes and Current Limitations

• The app currently loops forever until manually stopped (Ctrl+C).
• assistant.mp3 is written on each response and not deleted automatically.
• The assistant is intentionally restricted to pizza-ordering conversations.
• First-time Whisper model loading can take longer and may require additional ML runtime dependencies.

Troubleshooting

• Microphone not detected
  • Check OS microphone permissions and default input device.
• No audio playback
  • Confirm system audio output is working and pygame initialized correctly.
• API/auth errors
  • Verify .env keys are valid and loaded.
• Whisper/ML dependency issues
  • Ensure all requirements are installed in the active virtual environment.

Future Improvements

• Add text-input fallback mode
• Add graceful exit command (for example: "quit")
• Delete temporary audio files after playback
• Add unit and integration tests
• Package as a CLI application