Interactive Sound Event Detector (I-SED)

I-SED is a human-in-the-loop interface for sound event annotation that helps users label sound events of interest within a lenghty recording quickly. The system speeds up human’s searching for a set of sound events of interest in an audio recording when there are too few labeled examples (e.g., one) of the sound class to train a state-of-the-art machine audio labeling system.

Demo video

You can also try the demo interface on your local machine: https://github.com/bongjun/ised

Talk (IUI 2019)

Related papers:

• [pdf] Bongjun Kim and Bryan Pardo, “A Human-in-the-loop System for Sound Event Detection and Annotation,” ACM Transaction on Interactive Intelligent System (TiiS), Vol. 8, Issue 2, Article 13, July 2018.

• Bongjun Kim, “Leveraging User Input and Feedback for Interactive Sound Event Detection and Annotation,” Doctoral Consortium, ACM International Conference on Intelligent User Interfaces (IUI) 2018

• Bongjun Kim and Bryan Pardo, “I-SED: an Interactive Sound Event Detector,” ACM International Conference on Intelligent User Interface (IUI), Limassol, Cyprus, Mar. 2017

• Bongjun Kim and Bryan Pardo, “Interface Design for Interactive Sound Event Detection,” Workshop on Awareness Interface and Interaction (AWARE) at the ACM International Conference on Intelligent User Interface, Limassol, Cyprus, Mar. 2017