added test for websocket

reduced threshold for alert system

updated the title :)

load model from huggingface instead of local

readme updated

added 60 sec limit

improved the streamlit ui

added alerts for the ui

improved the real time detection

websockets for realtime added

updated the streamlit ui, nvm =)

updated the streamlit ui to include a microphone

improved feature extractor with normalization and caching

streamlit ui added

dependencies updated

changed fastapi script to use onnx int8

added inference script using pytorch model

added benchmark script to test models

added onnx conversion script

complete fine-tuning pipeline and inference engine

added a model evaluation script

readme updated

implement Wav2Vec2 feature extraction and bulk processing

initialize project structure and data harmonization pipeline