Creating a real-time, 100% offline, low-power environment monitor that translates ambient life-saving sounds into haptic/visual feedback for the hearing impaired. /Dhwani-AI │ ├── /models # Pre-trained and Quantized (ONNX/Vitis-AI) models ├── /src # Python/C++ code for NPU-accelerated inference │ ├── audio_stream.py │ ├── amd_npu_engine.py ├── /frontend # Dashboard/Mobile App code ├── /data # Sample sound datasets for testing └── README.md # Documentation, AMD setup guide, and USP