Can a machine “listen,” and can it “understand” what it listens?
The installation The Listening Machine is not about “understanding” in the comprehensive sense — that would be too big and complex a field. But it shows that a machine can learn to correctly recognize and classify sounds from its surroundings that it hears through its “ear,” i.e. the microphone. An artificial neural network was trained on thousands of sound samples to be able to differentiate various kinds of sounds. In the process, characteristic patterns in the audio signal are determined for the different sounds, for example, what differentiates a flute from a trumpet or spoken language from singing. The technology behind it is machine learning, but some simply call it statistics.
Credits: Stefan Balke, Matthias Dorfer, Florian Henkel, Alexander Moser, Andreas Arzt, and Gerhard Widmer: Institute of Computational Perception, Johannes Kepler University (JKU) Linz, www.cp.jku.at; LIT | AI Lab, Linz Institute of Technology (LIT), Linz, www.jku.at/en/linz-institute-of-technology/research/research-labs/artificial-intelligence-lab. This research has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 670035 “Con Espressione”).