On the use of U-Net for dominant melody estimation in polyphonic music
In: MMRP 2019 ; https://hal.science/hal-02457728 ; MMRP 2019, Jan 2019, Milano, Italy, 2019
Online
Konferenz
Zugriff:
International audience ; Estimation of dominant melody in polyphonic music remains a difficult task, even though promising breakthroughs have been done recently with the introduction of the Harmonic CQT and the use of fully convolutional networks. In this paper, we build upon this idea and describe how U-Net-a neural network originally designed for medical image segmentation-can be used to estimate the dominant melody in polyphonic audio. We propose in particular the use of an original layer-by-layer sequential training method, and show that this method used along with careful training data conditioning improve the results compared to plain convolutional networks.
Titel: |
On the use of U-Net for dominant melody estimation in polyphonic music
|
---|---|
Autor/in / Beteiligte Person: | Doras, Guillaume ; Esling, Philippe ; Peeters, Geoffroy ; Analyse et synthèse sonores Paris ; Sciences et Technologies de la Musique et du Son (STMS) ; Institut de Recherche et Coordination Acoustique/Musique (IRCAM)-Sorbonne Université (SU)-Centre National de la Recherche Scientifique (CNRS)-Institut de Recherche et Coordination Acoustique/Musique (IRCAM)-Sorbonne Université (SU)-Centre National de la Recherche Scientifique (CNRS) ; Représentations musicales (Repmus) ; Institut Polytechnique de Paris (IP Paris) ; Laboratoire Traitement et Communication de l'Information (LTCI) ; Institut Mines-Télécom Paris (IMT)-Télécom Paris ; Signal, Statistique et Apprentissage (S2A) ; Institut Mines-Télécom Paris (IMT)-Télécom Paris-Institut Mines-Télécom Paris (IMT)-Télécom Paris |
Link: | |
Zeitschrift: | MMRP 2019 ; https://hal.science/hal-02457728 ; MMRP 2019, Jan 2019, Milano, Italy, 2019 |
Veröffentlichung: | HAL CCSD, 2019 |
Medientyp: | Konferenz |
Schlagwort: |
|
Sonstiges: |
|