Cognitive system that hears more objectively than humans

The area of testing machines and products by means of acoustic signals is a young but growing field in industrial production. Ensuring that machines are operating properly is a continuous process and one of the ways this is done is through the sounds the machine makes. People are usually involved in this process to listen out for abnormal sounds indicating problems that need to be corrected.

However, since everyone perceives sound somewhat differently, detection of problems may be rather subjective and may involve a higher risk of error. To compensate for this, the Fraunhofer Institute for Digital Media Technology (IDMT) has developed cognitive systems that accurately identify faults based on acoustic signals. Initial practical tests show error detection rates of up to 99%. The technological approach combines intelligent acoustic measurement technology, signal analysis, machine learning and data-safe, flexible data storage, reports Phys.org.

“We integrate the intelligence of listening into the industrial condition control of machines and automated test systems for products,” explains Steffen Holly of IDMT’s Industrial Media Applications business unit. The idea is that cognitive systems can hear more objectively than humans once they have been trained, because they have millions of neutral data records at their disposal.

Source: Phys.org.