AES Berlin 2017
Poster Session EB01
EB01 - Posters: Analysis, Coding, and Hearing
Saturday, May 20, 15:00 — 18:00 (Gallery Window Area 2)
EB01-01 Source Separation in Action: Demixing the Beatles at the Hollywood Bowl
James Clarke (Presenting Author)
Except for unamplified dramatic performances or concerts where the audience is either used to sitting quietly or can be persuaded to listen quietly, high levels of amplifications are required so that the speech and music can still be heard above the self-noise generated by the crowd. Crowd noise can reach surprisingly high sound levels when several thousand people are shouting at each other, all at the same time. In this paper a method is presented that was used to isolate the crowd noise non destructively, so the raw instrumentation can be targeted in isolation. These isolated sources now become available for re-mixing and balancing.
Engineering Brief 307
EB01-02 Feature Selection for Real-Time Acoustic Drone Detection Using Genetic Algorithms
Marta Bautista-Durán (Presenting Author), Joaquin García-Gomez (Author), Roberto Gil-Pita (Author), Manuel Rosa-Zurera (Author)
Drones are taking off in a big way, but people sometimes use them in order to invade the privacy of others or to bypass the security systems, making their detection an actual issue. The objective of the proposed system is to design real-time acoustic drone detectors, able to distinguish them from objects that can be acoustically similar. A set of features related to the propeller sounds have been extracted, and genetic algorithms have been used to select the best subset. The classification error achieved with 30 features is below 13%, making feasible the real-time implementation of the proposed system.
Engineering Brief 308