A real-time voice-based chatbot that listens to your voice, understands your question using LLaMA 2, and talks back using text-to-speech — creating an interactive assistant experience.
- 🎙️ Takes input from:
- Microphone
- Webcam (audio)
- OR Manual text
- 🎧 Converts voice to text using OpenAI Whisper
- 🧠 Understands your message using Meta LLaMA 2 (via Hugging Face Transformers)
- 🗣️ Speaks the response using pyttsx3 (text-to-speech)
- 🪄 Feels like you’re talking to a smart assistant!
| Function | Tool/Library |
|---|---|
| Speech Recognition | SpeechRecognition, Whisper |
| Natural Language Model | LLaMA 2 via transformers |
| Text-to-Speech | pyttsx3 |
| Audio Input Handling | OpenCV, pyaudio |
| Model Hosting | Hugging Face |
| Interface | Python CLI |
⚠️ This project must be run locally — microphones and webcams are not supported in Google Colab. However, text mode will still work there.
ℹ️ This is a work in progress and not yet fully optimized.
Copy and paste this into your terminal:
pip install SpeechRecognition
pip install soundfile
pip install opencv-python-headless
pip install torch
pip install transformers
pip install huggingface_hub
pip install pyttsx3
# OS-specific dependencies (for TTS and mic access)
sudo apt install espeak
sudo apt-get install portaudio19-dev
pip install pyaudio