On the use of U-Net for dominant melody estimation in polyphonic music - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

On the use of U-Net for dominant melody estimation in polyphonic music

Résumé

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.
Fichier principal
Vignette du fichier
On the use of U-Net for dominant melody estimation.pdf (482.21 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02457728 , version 1 (28-01-2020)

Identifiants

  • HAL Id : hal-02457728 , version 1

Citer

Guillaume Doras, Philippe Esling, Geoffroy Peeters. On the use of U-Net for dominant melody estimation in polyphonic music. MMRP 2019, Jan 2019, Milano, Italy. ⟨hal-02457728⟩
68 Consultations
255 Téléchargements

Partager

Gmail Facebook X LinkedIn More