Encouraging accessible conversations using sound scene monitoring and IoT-based nudging
Older adults with hearing loss often find it hard to follow group conversations in social contexts, especially when there is loud background noise and reverberation. Not being able to engage leads many people to opt out all together, which can have devastating consequences for mental health and increases the risk of dementia. This project will prototype solutions that can detect when conversations are becoming difficult to follow and help the group to keep everyone engaged.
About the project
The impact of hearing loss can be devastating, both for the individual and their family and friends. Hearing aids can help by amplifying quiet sounds enough to be heard, which works well for watching TV and having a one-to-one conversation in a quiet place. However, many everyday environments are inherently noisy. Hearing aid technology is advancing all the time, but it is not yet sufficient to solve the “speech-in-noise” problem. Struggling to follow a group conversation in a café or restaurant is frustrating and leads to a sense of social exclusion. This can often result in an individual opting out of social situations altogether, causing isolation, loneliness and even accelerating the onset of dementia.
The project aims to develop technology-assisted strategies which will help hearing impaired people to fully participate in group conversations. The difficulty of hearing well in a listening situation is affected by several factors, including the acoustics of the room, the locations of noise sources, the spatial configuration of the conversation participants and their group dynamics. The project will create easy-to-use tools which enable anyone to choose a good venue, sit in the best seat and provide real-time assistance to the group to help them talk in a manner which is accessible to the hearing-impaired person.
The project will exploit recent advances in machine learning and readily-available Internet-of-Things devices to understand sound scenes. Through user engagement and rapid design iterations it will develop appropriate “nudge” interventions which subtly guide the conversation participants to adopt more inclusive behaviour.