F. J. Canadas-Quesada, J. J. Carabias-Orti, R. Mata-Campos, N. Ruiz-Reyes, and P. Vera-Candeas, “Polyphonic Piano Transcription Based on Spectral Separation,” in Proc. AES Convention 124, May 2008, Paper 7384. [Online]. Available: https://aes.org/publications/elibrary-page/?id=14514
Canadas-Quesada FJ, Carabias-Orti JJ, Mata-Campos R, Ruiz-Reyes N, Vera-Candeas P. Polyphonic Piano Transcription Based on Spectral Separation. In: AES Convention 124. Audio Engineering Society; 2008. Paper 7384. Available from: https://aes.org/publications/elibrary-page/?id=14514
@inproceedings{CanadasQuesada2008_14514,
author = {Canadas-Quesada, Francisco Jesus and Carabias-Orti, Julio Jose and Mata-Campos, Raul and Ruiz-Reyes, Nicolas and Vera-Candeas, Pedro},
title = {{Polyphonic Piano Transcription Based on Spectral Separation}},
booktitle = {AES Convention 124},
note = {Paper 7384},
year = {2008},
month = may,
publisher = {Audio Engineering Society},
url = {https://aes.org/publications/elibrary-page/?id=14514}
}
TY - CPAPER
TI - Polyphonic Piano Transcription Based on Spectral Separation
AU - Canadas-Quesada, Francisco Jesus
AU - Carabias-Orti, Julio Jose
AU - Mata-Campos, Raul
AU - Ruiz-Reyes, Nicolas
AU - Vera-Candeas, Pedro
T2 - AES Convention 124
M1 - Paper 7384
PY - 2008
DA - 2008/05/06
UR - https://aes.org/publications/elibrary-page/?id=14514
PB - Audio Engineering Society
LA - en
AB - We propose a discriminative model for polyphonic piano transcription. Spectral features are obtained individually for each note. To solve the overlapping partial problem, we apply spectral separation by estimating the spectral envelope for each note. For classifying purposes, support vector machines (SVM) are trained on the spectral energy inferred from these spectral features. We apply a scheme of one-versus-all (OVA) SVM classifiers to discriminate frame-level note instances. To decrease the high frequency notes residual energy due to the downward notes shared partials, a method to cancel the interferences from the downward notes to the upward notes has been developed. The classifier output is filtered with a hidden Markov model. Our approach has been tested with synthesized and real piano recordings obtaining very promising results.
ER -
