Computers that listen could make sound valuable


Sound often is the clue that leads us to a machine’s malfunction. A mechanic might hear that a factory unit is working poorly, or a driver might hear a problem with a vehicle’s brakes. Now a slate of start-ups is betting that computers too can learn to diagnose mechanical problems based on sound, as Steven Melendez reports for FastCompany. One founder says that acoustic data are essentially neglected. Companies potentially could use them to monitor remote industrial equipment or even automatically diagnose problems in consumer products. But it’s not just a matter of collecting the data – the key, Melendez reports, is using machine learning to extract meaning from audio.

Read more in FastCompany >