DIN Industrial Noise Corpus

This corpus contains high-quality field recordings of industrial machine noise collected for research in speech recognition and human–machine interaction (HMI) within challenging industrial environments. The dataset was developed as part of the Arrowhead Project, an EU-funded initiative aimed at improving the competitiveness and efficiency of European manufacturing through advanced digital technologies.

As manufacturing industries face increasing production costs, there is a growing need for more intuitive and intelligent methods of interaction between operators and machines. The Arrowhead Project addresses this challenge by supporting the development of voice-controlled and sound-aware systems that can operate effectively in real-world industrial settings.

To support this goal, the SDU Acoustics Laboratory at the University of Southern Denmark has established a comprehensive database of authentic industrial machine sounds recorded directly in operational environments. These recordings provide realistic acoustic conditions for the development, training, and validation of speech recognition, audio event detection, and human–machine interaction technologies.

The open acoustic corpus serves as a valuable resource for researchers and engineers working in industrial acoustics, noise characterization, speech enhancement, machine listening, and intelligent manufacturing systems. By providing access to real-world industrial soundscapes, the dataset helps advance the development of robust audio-based technologies for future smart factories.