MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202641048360 A) filed by Saveetha Engineering College on April 16, 2026, for Ai-Based Emergency Keyword Detection System For Womens Safety.
Inventors include N. Madhumitha; Prashanth K; and Stephen Raj Y.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: Women's safety remains a top social priority, particularly in situations where immediate assistance is needed. However, traditional safety mechanisms such as manual SOS apps and physical panic buttons often fail during emergencies, as the victim may not be able to access or operate them. To address this challenge, we propose an intelligent, hands-free AI-based Emergency Keyword Detection System that can provide rapid support without any manual intervention. This system is specifically designed to enhance women's security both in public areas and within domestic environments. The solution continuously monitors live audio input and uses machine learning to detect a predefined emergency keyword- "Help". When spoken, even in tense conditions, the system processes the voice signal in real time. It extracts speech features using MFCCs (Mel-Frequency Cepstral Coefficients), converting short audio segments into meaningful numerical representations. These features are then fed into a Random Forest classifier, which has been trained on samples of the keyword and surrounding background noise to accurately differentiate between actual distress commands and regular sounds. Once the keyword is identified, the system automatically initiates a rapid emergency response. An alert workflow is instantly triggered, sending out an emergency SMS and placing a phone call to preconfigured contacts or authorities. The integrated geolocation module attaches the user's real-time location, enabling swift response and timely rescue actions. This approach minimizes human dependency and ensures that help can be summoned even when physical access to a device is restricted. The system is highly accessible, as it requires no special hardware and can be deployed easily on common smartphones and connected devices using cloud services. Testing under varied background noise conditions proves the robustness and reliability of the model, making it practical for real-world use. This technology stands as a low-cost, scalable and effective safety tool, with potential for integration into mobile apps, wearables, and web platforms for wider reach. By providing seamless, voice-activated emergency support, the model contributes significantly toward enhancing women's safety and building a more responsive protection ecosystem.
Disclaimer: Curated by HT Syndication.
