All projects
Project Study
Erradisnore
Embedded Snoring Detection and Positional Therapy Device
Embedded Systems
TinyML
Biomedical Engineering
Sensor Fusion
Erradisnore is an embedded system designed to detect snoring and identify supine sleeping position in real time. When repeated snoring events occur while the user is lying on their back, the device activates vibration feedback to encourage a change in sleeping position.
System Overview
Architecture
- 01Microphone
- 02TinyML Snoring Detection
- 03Snoring Event Logic
- 04IMU Posture Detection
- 05Supine + Repeated Snoring
- 06Vibration Feedback
Technical Details
My Contribution
- Embedded firmware development in C/C++
- TinyML audio classifier deployment
- Snoring event detection logic
- Integration of Arduino Nicla Voice
- Syntiant NDP120 inference
- BMI270 IMU integration
- Supine posture detection
- Vibration motor control
- Event buffering and storage logic
- Git-based version control
Embedded Machine Learning
On-device classification
The audio classifier was trained to distinguish snoring events from non-snoring audio and deployed directly on embedded hardware.
- On-device inference
- No continuous cloud processing required
- Real-time event detection
Model result
Approx. 86% model accuracy
Technical Toolkit
Technologies
C++
Arduino
Edge Impulse
TinyML
Syntiant NDP120
BMI270
IMU
Git
