DIGITAL

Let’s teach computers to listen! 20 years of AI in acoustic research and development at JOANNEUM RESEARCH

Publication from Digital
Intelligente Akustische Lösungen

Franz Graf, Clemens Amon , Martin Blass, Moritz Fišer, Ferdinand Fuhrmann, Florian Krebs

Elektrotechnik & Informationstechnik (2021) 138/3: 139–147 https://doi.org/10.1007/s00502-021-00882-5, 3/2021

Abstract:

Hearing, perception and interpretation of sound is natural to humans but has been regarded an imaginable task for computers and machines starting with the middle of the past century. Within the past decades, breaking technologies in the field of acoustics, audio engineering and computer sciences paved the way for teaching computers to listen and to analyze sound in real time - just like humans do. In this article, we would like to share our experiences as a research group, which has been actively working in the field of acoustic research and development with focus on acoustic monitoring over the past 20 years. We explain the necessary steps in implementing machine learning systems based on audio data, give insights about common challenges from the past and presence, and present application examples where acoustic intelligence solves tasks in fields like traffic, industry, safety and security.