HER SHIELD: Speech Emotion Recognition for Women’s Safety

Women’s safety remains a critical concern, requiring innovative technological solutions to provide timely assistance in emergencies. This paper presents Her_Shield, a speech-based emotion detection system designed to identify distress and trigger alerts. The system leverages a fine-tuned Whisper v3 model for speech-to-text conversion and sentiment analysis, detecting fear in real time. Additionally, a keyword detection mechanism enables hands-free activation using voice commands like "Activate Shield" and "Stop". Upon detecting distress, the system retrieves the user’s GPS location and sends emergency alerts via an SMS gateway to predefined contacts. By integrating machine learning, natural language processing, and geolocation tracking, Her_Shield aims to enhance personal safety with minimal user intervention..