Auralis

Auralis—Detect the decline, while there’s still time.

Comment 1

Auralis: Speech-Based Detection of Cognitive Decline

Every 3 seconds someone in the world develops dementia. With the stats expecting it to affect 1 in 10 people by 2050, this growing crisis will profoundly impact us, our friends, and our families. Auralis aims to contribute to early diagnosis and care by providing a simple yet powerful tool for assessing cognitive decline through direct speech.

Auralis is an innovative solution designed to detect early signs of dementia by analysing cognitive decline in real-time speech. Dementia, which is often caused by neurodegenerative diseases like Alzheimer’s and Parkinson’s, manifests through shared symptoms such as slower speech, long pauses, verbal fillers, and declined vocabulary. Early detection of these symptoms is key to timely medical intervention enabling dementia patients and their families to plan, find support, and prioritise what truly matters.

For this project, we combined advanced techniques in natural language processing with recent research on speech biomarkers in neurodegeneration, utilising the Hugging Face framework and integrating our machine learning model with the developed interface.

We used DementiaNet, an open-source dataset of audio recordings from famous actors with and without dementia. These recordings include speech patterns captured over time, allowing our machine-learning model to differentiate between healthy ageing and early cognitive decline.

Our user-friendly web application, Auralis, enables users to record 30-second to 1-minute audio clips for analysis, integrating into regular health routines. The model processes these recordings to identify dementia-related speech patterns for early detection. The app's intuitive design and ease of accessibility make it valuable for both clinicians and individuals to proactively monitor cognitive health.

Team: BrainSentry - Problem Division

Share this project:

Updates