chatbot-encodedmind 1.0.4

Voice-based AI chatbot powered by OpenAI

Chatbot

This project is a voice-based chatbot that uses OpenAI's API to interact with users. The chatbot listens to user input via voice, processes the input using OpenAI's whisper model for transcription, generates a response using the GPT-3.5-turbo model, and replies using text-to-speech model tts-1.

---

Features

• Voice Input: Users interact with the chatbot using their voice.
• Real-Time Transcription: Converts voice input into text using OpenAI's whisper model.
• Intelligent Responses: Generates context-aware and empathetic responses using OpenAI's GPT-3.5-turbo model.
• Voice Output: Uses OpenAI's tts-1 model to respond back in audio form.
• Continuous Conversation: Maintains conversation history for context-aware interactions.
• Language support: whisper can detect any language. Your accent can affect language detection!

---

Prerequisites

• Python 3.9+
• uv — a modern Python package manager
• OpenAI API Key
• Virtual environment (optional but recommended)
• A working microphone and speaker

---

Installation

Option 1: Install from PyPI

pip install chatbot-encodedmind

Option 2: Install from source

1. Clone the repository

   git clone https://github.com/EncodedMind/Chatbot.git
   cd Chatbot
   

2. Create and activate a virtual environment

   python -m venv venv
   source venv/bin/activate # On Windows: venv\Scripts\activate
   

3. Sync dependencies

   uv sync
   

4. Set your OpenAI API key Replace YOUR_OPENAI_API_KEY in the script with your actual OpenAI API key.

---

Usage

1. Run the chatbot If installed via PyPI:

chatbot

or, if installed from source:

uv run chatbot

2. Interact with the chatbot

• Speak into your microphone when prompted.
• Listen to the chatbot's response.
• Say "Goodbye" to end the session and exit the program.

---

Project Structure

├── chatbot/
│   ├── __init__.py
│   ├── ai_client.py       # Handles OpenAI API calls
│   ├── audio_io.py        # Manages audio input/output
│   └── core.py            # Main conversation loop
├── pyproject.toml         # Project metadata and dependencies
├── .python-version        # Python version for uv
└── README.md              # Project documentation

---

Key Components

Functions

• record_audio(duration, fs, device): Captures audio input from the user.
• get_chatgpt_response(conversation_history): Sends user messages to the GPT-3.5-turbo model and retrieves responses.
• main(): Orchestrates the chatbot's voice input/output loop and conversation logic.

Libraries Used

• OpenAI: For GPT and Whisper APIs.
• Sounddevice: For capturing audio input.
• Numpy: For handling audio data.
• Pydub: For playing audio responses.
• Scipy: For saving audio files as .wav.
• Requests: For making HTTP requests to the OpenAI API.

---

Future Improvements

• Silence Detection: Avoid the limitation of defining a duration parameter.

---

Notes

Ensure you have a stable internet connection for API calls. The interaction pause time (or response speed) of the chatbot depends on the quality and speed of your internet connection, as it relies on real-time API calls to OpenAI servers. Modify the chatbot behavior by changing the initial conversation_history system message.

---

Troubleshooting

Audio Issues: Ensure your microphone is properly configured and accessible to the program. API Errors: Verify your API key and check for usage limits on your OpenAI account. Dependency Issues: Ensure all required libraries are installed using uv sync.

---

Project Extension: Teddy Bear Voice Assistant

In addition to the core functionality of the voice chatbot, I extended the project by integrating it with an Asus Tinker Board. The board acts as the central controller for the chatbot's operations. I connected a microphone and speaker to the Tinker Board and embedded everything inside a teddy bear to create a kid-friendly interactive toy, similar to the Furby.

This project extension aims to provide a fun and engaging experience for children while maintaining the capabilities of the voice chatbot.

Watch the Interaction in Action

You can see the teddy bear in action with the chatbot by watching this video on YouTube:

• Video: Teddy Bear Voice Assistant Demo

---

License

This project is open-source and available under the MIT License.

---

Acknowledgements

• OpenAI for providing the models.
• The developers of Sounddevice, Pydub and Scipy for enabling audio processing.

---

Enjoy using the Chatbot!