MAIN MELODY EXTRACTION WITH SOURCE-FILTER NMF AND CRNN - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2018

MAIN MELODY EXTRACTION WITH SOURCE-FILTER NMF AND CRNN

Résumé

Estimating the main melody of a polyphonic audio recording remains a challenging task. We approach the task from a classification perspective and adopt a convolutional recurrent neural network (CRNN) architecture that relies on a particular form of pretraining by source-filter nonneg-ative matrix factorisation (NMF). The source-filter NMF decomposition is chosen for its ability to capture the pitch and timbre content of the leading voice/instrument, providing a better initial pitch salience than standard time-frequency representations. Starting from such a musically motivated representation, we propose to further enhance the NMF-based salience representations with CNN layers , then to model the temporal structure by an RNN network and to estimate the dominant melody with a final classification layer. The results show that such a system achieves state-of-the-art performance on the MedleyDB dataset without any augmentation methods or large training sets.
Fichier principal
Vignette du fichier
273_Paper.pdf (375.98 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02019103 , version 1 (14-02-2019)

Identifiants

  • HAL Id : hal-02019103 , version 1

Citer

Dogac Basaran, Slim Essid, Geoffroy Peeters. MAIN MELODY EXTRACTION WITH SOURCE-FILTER NMF AND CRNN. 19th International Society for Music Information Retreival, Sep 2018, Paris, France. ⟨hal-02019103⟩
497 Consultations
725 Téléchargements

Partager

Gmail Facebook X LinkedIn More