This repository contains the early-stage prototype for the "Voice-Driven Banking using LLMs" project under Mifos for C4GT 2025.
Leverage open-source STT (Speech-to-Text) and TTS (Text-to-Speech) tools to enable voice-driven banking experiences for underserved users via conversational interfaces powered by LLMs.
- Simulated voice input/output flow
- Placeholder for integration with Whisper/Vosk STT engine
- Simple conversational interface for mock banking queries
- Modular code for future LLM API integration (LLAMA2/OpenChat)
.
βββ README.md
βββ app.py # Main Flask app
βββ templates/
β βββ index.html # UI for the voice interaction demo
βββ static/
β βββ style.css # Basic styles
βββ voice/
βββ mock_stt.py # Simulated STT processor (placeholder)
- Integrate actual Whisper or Vosk STT module
- Add real-time TTS response (pyttsx3 or gTTS)
- Connect to sample banking APIs for real use-cases
- Add Hindi & Telugu language support
PRs and issues welcome! Built for the Mifos community with β€οΈ