AN EXEMPLAR-BASED NMF APPROACH FOR AUDIO EVENT DETECTION

We present a novel, exemplar-based method for audio event detection based on non-negative matrix factorisation (NMF). Building on recent work in noise robust automatic speech recognition, we model events as a linear combination of dictionary atoms, and mixtures as a linear combination of overlapping events. The exemplarbased dictionary is created by extracting all available training data, artificially augmented by linear time warping at multiple rates. The method is evaluated on the Office Live and Office Synthetic development datasets released by the AASP Challenge on Detection and Classification of Acoustic Scenes and Events.