Mid-level sparse representations for timbre identification: design of an instrument-specific harmonic dictionary

Pierre Leveau 1 Emmanuel Vincent 2 Gaël Richard 1 Laurent Daudet 3
1 Télécom ParisTech
2 Centre for Digital Music
3 DALEMBERT - Institut Jean Le Rond d'Alembert
Abstract : Several studies have pointed out the need of mid-level representations of music signals for information retrieval and signal processing applications. In this paper, we investigate a new representation based on sparse decomposition of the signal into a collection of instrument-specific harmonic atoms modelling notes of various pitches played by different instruments. Each atom is composed of windowed harmonic sinusoidal partials whose amplitudes are learned on a training database. An efficient Matching Pursuit algorithm was designed to extract the best atoms and to estimate the phases of their partials. Then we explain how the resulting representation can be exploited for automatic instrument recognition. Preliminary experiments on a test database of solo excerpts show promising results.
Type de document :
Communication dans un congrès
1st Workshop on Learning the Semantics of Audio Signals (LSAS), Dec 2006, Athens, Greece. 2006
Liste complète des métadonnées

Littérature citée [14 références]  Voir  Masquer  Télécharger
https://hal.inria.fr/inria-00544284
Contributeur : Emmanuel Vincent <>
Soumis le : mardi 7 décembre 2010 - 15:54:40
Dernière modification le : jeudi 22 novembre 2018 - 14:25:50
Document(s) archivé(s) le : mardi 8 mars 2011 - 04:42:09

Fichier

leveau_LSAS06.pdf
Fichiers éditeurs autorisés sur une archive ouverte

Identifiants

  • HAL Id : inria-00544284, version 1

Collections

UPMC | IJLRDA | SORBONNE-UNIVERSITE

Citation

Pierre Leveau, Emmanuel Vincent, Gaël Richard, Laurent Daudet. Mid-level sparse representations for timbre identification: design of an instrument-specific harmonic dictionary. 1st Workshop on Learning the Semantics of Audio Signals (LSAS), Dec 2006, Athens, Greece. 2006. 〈inria-00544284〉

Exporter

BibTeX TEI DC DCterms EndNote

Partager

Métriques

Consultations de la notice

142

Téléchargements de fichiers

140

Contact


archive-ouverte@inria.fr