First prize · Hack4Change 2026
SirenSense is a three-device road-awareness prototype designed to make emergency vehicles easier to notice, especially for deaf and hard-of-hearing drivers. It combines siren detection and direction finding with cameras, vehicle sensors, navigation and an emergency-vehicle transponder.
Why it exists
Emergency sirens are easy to miss inside a modern car, and an audio-only warning excludes drivers who cannot hear it. SirenSense turns surrounding sound and vehicle context into visible, spoken and directional information.
Not one sensor. A complete safety ecosystem.Designed and built by Ibrahim Najjar
System architecture
The roof node listens outside the vehicle. The cabin node combines those results with cameras and onboard sensors. The transponder lets an emergency vehicle broadcast authenticated position data to nearby prototypes.
A weather-resistant prototype built around one ESP32-S3 and four digital MEMS microphones arranged on a 190 mm cross.
An ESP32-S3 safety controller paired with a Raspberry Pi interface for alerts, cameras, navigation, sensing and logging.
A separate demo unit using GPS and 433 MHz LoRa to share an emergency vehicle's position with nearby SirenSense systems.
More than siren detection
Features are marked by their real project status. “Implemented” means present in the prototype or working software. “In integration” still needs complete vehicle testing. “Concept” is not presented as a certified product function.
A TinyML audio model trained with real roadside recordings, including traffic, wind and Doppler effects.
Implemented prototypeGCC-PHAT time-delay estimation turns four microphone streams into a left, right, front or rear bearing.
Implemented prototypeFront camera capture and an interface designed for Full HD recording at 30 frames per second.
Software implementedA driver-facing camera pipeline for fatigue, distraction and phone-use warnings.
In integrationForward radar data feeds distance and closing-risk alerts into the cabin interface.
In integrationAir-quality, humidity and temperature measurements support driver information and device protection.
Integrated sensingAn accelerometer and gyroscope support collision events and driving-motion records.
Integrated sensingA custom vector-map interface provides GPS navigation without depending on a cloud map service.
Working softwareThe Raspberry Pi interface includes OBD-II gauges and system diagnostics alongside safety alerts.
Software implementedA prototype emergency mode shares position with nearby SirenSense units and activates the roof light.
Concept · not DGT certifiedLoRa and GPS add a direct position channel from participating emergency vehicles.
Demo unit builtBMS, charging, DC/DC conversion, filtered rails, decoupling and device-temperature monitoring are part of the design.
Implemented prototype
The four-microphone solution
The INMP441 can select the left or right slot of an I²S stereo frame. That allows each opposing microphone pair to share one data line. East and west use I²S0; north and south use I²S1. I²S0 supplies BCLK and WS, and the ESP32-S3 GPIO matrix routes those clocks internally to I²S1 so both pairs stay aligned.
Roof-node power path
The less visible engineering
The microphones, radio and converter share a small enclosure, so power integrity is part of the sensing problem. The roof node includes battery protection, LiFePO₄ charging through a TP5000 module, boost conversion, bulk and ceramic capacitance, an LC filter and a ferrite bead at every microphone branch.
Engineering status
SirenSense is a research and demonstration prototype, not a homologated automotive safety product. The site deliberately avoids turning design estimates into measurements.
Four-microphone capture on one ESP32-S3, siren classification, directional processing, ESP-NOW messages, physical PCBs, enclosures and the award demonstration.
The cabin PCB has been fabricated and its firmware written. Final assembly, Raspberry Pi 5 vehicle testing, camera validation and complete sensor calibration remain ongoing.
Urban LoRa range and potential response-time reductions are engineering estimates until field-characterised. The enclosure is designed for weather resistance but has no IP certification.
Hack4Change 2026
SirenSense won first place at the Hack4Change event hosted at the University of Seville's ETSII. The award recognised the complete ecosystem: accessible road-safety concept, embedded hardware, software, communication between nodes and a working physical demonstration.
Inspect the project on GitHub ↗Go deeper