A terminal‑based AI assistant that supports both text and voice interaction in Indian regional languages. Current pipeline:

User (voice/text, regional) → Whisper (ASR) → Deep‑Translator → GPT‑3.5/4 → gTTS (TTS) → Audio/Text reply

---

✨ Features

---

🗂️ Project Structure

voicebot/
├── main.py            # Orchestrator with conversation loop
├── input_handler.py   # Handles text/voice input + language detection
├── transcriber.py     # Whisper transcription + mic recording
├── translator.py      # Regional → English translation (deep‑translator)
├── llm_client.py      # OpenAI GPT client (v1 SDK)
├── tts.py             # gTTS speech synthesis + playback
├── utils.py           # Helper (audio playback)
└── requirements.txt

---

🔧 Tools & Libraries

---

⚙️ Installation

# Create and activate venv
python3 -m venv voicebot-env
source voicebot-env/bin/activate

# Install Python dependencies
pip install -r requirements.txt

# System packages (Ubuntu)
sudo apt update
sudo apt install ffmpeg portaudio19-dev python3-pyaudio

requirements.txt contents:

openai
whisper
torch
sounddevice
scipy
deep-translator
langdetect
gTTS

---

🔑 Environment Variables

Set your OpenAI key once:

export OPENAI_API_KEY="sk-..."

---

🚀 Usage

python3 main.py
# Choose input mode: text or voice

• Exit any time by typing exit (text mode) or saying the word “exit” (voice mode).

---

🛣️ Roadmap

1. Replace gTTS with higher‑quality Kannada TTS (Google Cloud TTS / Bhashini)
2. Persist memory across sessions (simple JSON storage)
3. Add transcript logging per session
4. Optional GUI (Gradio) or Flask API wrapper