Voice Banking is a cutting-edge application designed to simplify everyday banking operations through voice commands in your local language. It leverages state-of-the-art technologies to deliver a seamless, user-friendly, and multilingual voice banking experience.
- Perform common banking operations like account balance inquiries, fund transfers, and transaction history checks.
- Support for multiple local languages for better accessibility.
- Advanced voice recognition powered by Whisper and IndicWhisper for high accuracy.
- Context-aware responses through Prompt Engineering and LLama 3 for natural conversations.
- Cross-platform compatibility built using Flutter.
- Python: Core backend logic and integration with machine learning models.
- Flutter: Frontend framework for building a responsive, cross-platform user interface.
- LLama 3: Language model for generating context-aware and natural conversations.
- Whisper & IndicWhisper: High-performance voice recognition systems optimized for multilingual support.
- Prompt Engineering: Fine-tuned prompts to improve interaction quality and contextual accuracy.
cd stt_service pip install -r requirements.txt uvicorn app:app --reload --port 8000
Test with curl(make sure you have the sample.wav file): curl -F "audio=@sample.wav" -F "session_id=test1" http://localhost:8000/voice/transcribe-intent