Researchers build AI to predict violin sound from its design

A team of researchers from Politecnico di Milano developed an artificial intelligence system capable of helping instrument makers determine how a handcrafted violin might sound.

The scientists’ algorithm evaluates designs, processes measurements, and compares them with the resulting vibrations. The network’s accuracy in predicting the sounds of instruments reached 98%, the researchers said.

According to them, violin makers still rely on tradition rather than understanding how the instrument’s shape affects its sound. The scientific approach could significantly enhance a craftsman’s ability to think, design, implement, and create, the scientists say.

“The ability to predict how a violin sounds could truly change the rules of the game for violin makers, as it would help them not only achieve more than grandmasters, but also explore the potential of new designs and materials,” the study says.

The scientists added that their method could democratise instrument production for those who lack the resources and time to experiment with shapes, sounds and materials.

Earlier in May 2021, researchers at the University of Glasgow developed a machine-learning algorithm that turns a smartphone or laptop into a sonar device.

In April, MIT researchers created a robot that can see objects through walls using radio-frequency identification.

In February, Google unveiled a technology for measuring heart rate and breathing rate using a smartphone camera.

Subscribe to ForkLog news on Telegram: ForkLog AI — all the news from the world of AI!