sound.count. is developing an acoustic AI model for counting and classifying motorized traffic. Our technology is designed to integrate across a range of use cases — from traffic planning tools and smart city infrastructure to ITS platforms and software SDKs.
From Sound to Traffic Intelligence
A microphone records ambient audio near the road. No camera, no radar, no invasive hardware — just sound.
Our model processes the audio to detect, count, and classify passing motorized vehicles — and can acoustically distinguish internal combustion engines from electric vehicles.
Outputs include vehicle counts, classifications, and noise level statistics — structured data ready to feed into planning tools, dashboards, or third-party platforms.
Built Different
Audio is processed on-device. No cameras, no license plate recognition — no personal data ever leaves the sensor.
Inference runs on lightweight embedded hardware. No cloud connection required — simple deployment and full data sovereignty.
Acoustic sensing eliminates the need for radar arrays, camera systems, or inductive loops — a fraction of the hardware cost per installation.
One Technology, Many Use Cases
A lightweight alternative to fixed side radar sensors. Acoustic sensing makes deployment fast and flexible — suited for short-term studies, mobile monitoring, or sites where installing fixed hardware isn't practical.
Beyond counting, the model outputs noise statistics alongside traffic data — enabling adaptive street lighting, noise-aware urban planning, and traffic intelligence embedded in existing smart city hardware.
Sensors placed ahead of intersections feed live traffic patterns into signal control systems — enabling smarter timing decisions that reduce wait times and improve throughput. A natural fit for ITS platforms.
European cities are legally required to evaluate the effectiveness of traffic calming measures and low emission zones. Acoustic traffic monitoring provides the before/after data needed to demonstrate impact — without cameras or expensive infrastructure.
Who We Are
Machine learning engineer since 2010 with over a decade of experience in the ITS industry, where he built OCR systems for traffic enforcement cameras deployed worldwide. Currently developing AI models at an international identity verification company. sound.count. brings both worlds together: deep traffic domain knowledge and hands-on AI engineering.
LinkedInProfessional CAD engineer and longtime friend of Gernot — the two studied together and have collaborated for over twenty years. At sound.count., Roman leads hardware design and enclosure engineering, and is actively engaged in business development through the aws First Incubator program.