Logo des Repositoriums
 

Decompositions of 2D feature representations with applications to acoustic event detection

dc.contributor.authorCornaggia-Urrigshardt, Alessia
dc.contributor.authorKurth, Frank
dc.contributor.editorHorbach, Matthias
dc.date.accessioned2019-03-07T09:32:23Z
dc.date.available2019-03-07T09:32:23Z
dc.date.issued2013
dc.description.abstractIn this paper we present an automatic procedure for detecting suddenly occurring events in a given audio signal. In particular, we are interested in events which show a quasi-periodic behavior such as recurring hands-clapping or sounds of a person knocking on a resonating surface. We exploit the analogy that impulse-like events such as the ones we want to detect have with percussive components in music: they both appear as vertical lines in the spectrogram. This property allows us to adapt techniques from Music Information Retrieval to our scope. In particular we perform a vertical/horizontal decomposition of the spectrogram to emphasize these vertical lines. In a further step, we detect the positions of these lines and consequently of the events we are interested in with the help of a novelty curve. Finally, the periodicity plays an important role in the process of discarding peaks of the novelty curve coming from background noise or sounds which we do not want to detect.en
dc.identifier.isbn978-3-88579-614-5
dc.identifier.pissn1617-5468
dc.identifier.urihttps://dl.gi.de/handle/20.500.12116/20702
dc.language.isoen
dc.publisherGesellschaft für Informatik e.V.
dc.relation.ispartofINFORMATIK 2013 – Informatik angepasst an Mensch, Organisation und Umwelt
dc.relation.ispartofseriesLecture Notes in Informatics (LNI) - Proceedings, Volume P-220
dc.titleDecompositions of 2D feature representations with applications to acoustic event detectionen
dc.typeText/Conference Paper
gi.citation.endPage2867
gi.citation.publisherPlaceBonn
gi.citation.startPage2853
gi.conference.date16.-20. September 2013
gi.conference.locationKoblenz
gi.conference.sessiontitleRegular Research Papers

Dateien

Originalbündel
1 - 1 von 1
Lade...
Vorschaubild
Name:
2853.pdf
Größe:
510.01 KB
Format:
Adobe Portable Document Format