UPLO | Devpost

Inspiration

Many of us struggle with understanding understanding our own thoughts, emotions, and behaviors in our fast-paced world.

Traditional support systems such as friends, family and therapists offer limited support and often lack real-time, unbiased insights.

What it does

Imagine a FitBit or Whoop, but instead of taking physical metrics like heart rate or blood oxygen, it listens to everything you say and how you say it. Using an emotionally intelligent AI (like HUME.ai) it will extract information on your mood and wellbeing each day so you can track the highs and lows of life and start getting unbiased emotional support to gain understanding of your wellbeing.

It is a wearable microphone that is always listening. Its USP is the microphone uses AI to recognise its owner's voice and deletes all other voices in real-time to protect the privacy of all others. Alternatively an on-skin or bone conduction microphone could be used to isolate just the user's voice if processing and deleting other voices in real-time is deemed to be a GDPR risk.

With the captured data the user will then be able to track their wellbeing and mood over time and deliver insights such as spells of activity, anxiety, satisfaction, over-exertion and many other metrics. Using ML/AI to analyse the captured data you will also be able to detect warnings of upcoming burnout, cognitive decline, depression and much more.

How we built it

Used a XIAO-nRF52840 Sense Micro Controller as the BLE Microphone.

Flutter App developed that connects to the AI pendant. It makes live transcriptions of what it hears and creates memories with summaries of each conversation.

Designed the pendant in CAD using Fusion 360 and then used a Elegoo Saturn DLP 3D printer to print the body of the necklace. Used the shoelase from my left shoe as the necklace. Found random coin-shaped metal discs at home and integrated them into the design to give the necklace some bling.

Challenges we ran into

3D Print partially failed overnight as part of the wall section was too thin around the USB-C port - but still workable.
Have a small LI-ION battery (but need a soldering iron to connect) so just running off USB power for now.
Have a bug in the app which means transcriptions stop coming through sometimes (not sure exactly why).

I am yet to integrate the HUME.ai emotional analysis into the app. This is a pretty simple API with SDK available.