AI-Based System for Rapid Identification of Dangerous Sounds for Individuals with Hearing Impairments

Funding from the Ministry of Education and Science

Project title: AI-Based System for Rapid Identification of Dangerous Sounds for Individuals with Hearing Impairments

Agreement number: DWD/8/0285/2024

Project implementation period: 01.10.2024 - 30.09.2028

Principal Investigator: dr hab. inż. Michał Wydra, profesor uczelni

Project value: 387 250,94 PLN

Funds granted for Lublin University of Technology: 387 250,94 PLN

Abstract: The project is carried out in the scientific discipline of technical computer science and telecommunications. The scientific goal of the project: to develop a method for rapid identification of sounds indicating danger, such as an ambulance siren, a fire alarm, a dog barking, a car horn, or a scream in the real environment of a person with hearing impairment using machine learning techniques and artificial intelligence algorithms. The method aims to identify and classify the threat based on the first phase of the sound (less than 0.25 seconds) despite noise or typical background sounds. The system combines advanced machine learning algorithms with mobile technologies and wearable devices, enabling real-time sound detection and user notification using vibrations, light notifications, and other sensory signals.

Current state of knowledge: Current technologies for supporting people with hearing loss and deafness are limited in their speed and effectiveness in responding to alarm sounds. To ensure immediate threat detection, existing systems often do not integrate advanced artificial intelligence (AI) algorithms and machine learning (ML) methods.

Justification for undertaking the research problem: The increase in the number of people with hearing problems emphasizes the need to develop technologies supporting their safety. There is an urgent need for a system that detects dangerous sounds at a speed comparable to the reactions of healthy people. There is a lack of integrated solutions that can be used on mobile and peripheral devices, such as smartwatches.

Justification for the innovative nature of the research: The AI ​​EAR SUPPORT project uses advanced AI and ML algorithms to detect alarm sounds quickly. The innovation lies in using neural networks trained to identify threats in various acoustic environments. The project assumes laboratory tests optimizing the system's response time by measuring the reaction times of healthy and hearing-impaired people.

Significance of the project results for developing a given field and scientific discipline: The AI ​​EAR SUPPORT project results can contribute to developing assistive technologies and AI. An effective alarm sound detection system will increase the safety of people with hearing problems and open up new research opportunities in AI integration in mobile and wearable applications. The developed algorithms can be adapted to other applications, such as security systems in smart cities or acoustic monitoring in industry. The AI ​​EAR SUPPORT project has the potential to become a breakthrough solution in technology supporting people with hearing loss and deafness. Its implementation will contribute to the progress in science and practical applications of AI, improving the quality of life of people with hearing problems.

Applications: The project's effects are of interdisciplinary importance, covering technical sciences and areas related to medicine and rehabilitating people with hearing disabilities. The system will find application in users' everyday lives, increasing their safety in public places, workplaces, and at home.

Expected results: Detecting dangerous sounds more efficiently to over 95% in complex acoustic conditions. Developing energy-saving AI algorithms adapted to mobile devices. Delivering a mobile application and server prototype ready for implementation in a real environment. The project is part of an implementation doctorate with an industrial technology partner, Freshware Sp. z o.o., which supports the development of the application and the integration of the system with mobile devices.