In June 2025, we traveled to Verona, Italy for a week-long collaboration with MoveByBike Italy, supported by CreativeFLIP and the European Commission. Our goal? To explore how sound design and machine learning could solve a real-world challenge: breakdowns of delivery bikes carrying heavy loads across rough, inner-city terrain.
MoveByBike operates hybrid cargo bikes—hauling up to 200kg—through narrow streets and cobblestone alleys where trucks can’t go. These innovative bikes are part of a growing industry, but mechanical stress is constant. We asked: Can we predict bike issues through sound before costly repairs are needed?
The result: a DIY acoustic monitoring system, built using Google’s Teachable Machine, that analyzes audio data to anticipate maintenance needs.
What’s more? We wanted to share what we learnt and release a step-by-step guide to help others build their own Acoustic Monitoring System to monitor the physical objects in the world around them using nothing but some cheap contact mics, a recorder and an internet-enabled computer. Visit the projects webpage to find out how: www.sonicassembly.se/work/velosonics/
Supported by Creative FLIP and @creativehubsnet @goetheinstitut.belgien
@EuropeanCommission
🎶 Original music by Joshua Ng.
#SoundDesign #MachineLearning #CargoBikes #CreativeTech #Verona #MoveByBike #SonicAssembly #DIYTech #UrbanMobility #TeachableMachine #SoundInnovation #CrossSectoralPioneers
MoveByBike is a micro logistics company
that operates in Verona It operates through only cargo bikes, especially in the city centre So coming from the world of audio I’ve always been interested in how sound
can be used to communicate information If you think about when you’re driving a car
for example, and you hear a weird noise and you know something’s wrong with it One driver comes in and tells us
“You know, I heard this sound” “But I don’t know where it’s coming from,
but it sounds so bad” Everything’s extremely subjective, and not organised learning The optimal scenario for us would be
several piezo sensors Strategically placed on the bike, so you can see from the different sections how many times can a bike go over a certain bump before it becomes an issue, for instance