Real-time Temporal Segmentation of Note Objects in Music Signals. Brossier, P., Bello, J. P., & Plumbley, M. D. In Proceedings of the International Computer Music Association, 2004.
  Title                    = {Real-time Temporal Segmentation of Note Objects in Music Signals},
  Author                   = {Brossier, P. and Bello, J. P. and Plumbley, M. D.},
  Booktitle                = {Proceedings of the International Computer Music Association},
  Year                     = {2004},

  Review                   = {Brossier \etal \cite{Brossier2004} want to segment musical notes via onset detection. In order to do so, the onset and offset of an isolated music note can be approximated by locating the Attack Decay Sustain Release characteristic waveforms. However, when notes are not played in isolation, this process is much more difficult. Several different methods to perform onset detection can be constructed. It can be facilitated by calculating the distance of the observed data's Short Time Fourier Transform (STFT) and that of a known note. Alternatively, the phase deviation or the spectral magnitude differences between two STFTs can be examined. In order to obtain a sequence of onset times, a peak-picking technique is used, where local maximas above a certain threshold is selected. Filtering and dynamic thresholding is used to improve the segmentation quality.},
  Timestamp                = {2013.08.29}

Downloads: 0