Onset Detection in Polyphonic Signals by means of Transient Peak Classification
Résumé
The article describes an onset detection algorithm that is based on a classification of spectral peaks into transient and non-transient peaks and a statistical model of the classification results to prevent detection of random transient peaks due to noise. This article describes the algorithmic changes compared to last years submission and discusses the conclusions drawn from the evaluation results.
Origine : Fichiers éditeurs autorisés sur une archive ouverte