Catégorie de document |
Contribution à un colloque ou à un congrès |
Titre |
Realtime Multiple Pitch Observation using Sparse Non-negative Constraints |
Auteur principal |
Arshia Cont |
Colloque / congrès |
International Symposium on Music Information Retrieval (ISMIR). Victoria : Octobre 2006 |
Comité de lecture |
Oui |
Année |
2006 |
Statut éditorial |
Publié |
Résumé |
In this paper we introduce a new approach for realtime multiplepitch observation of musical instruments. The proposed algorithm isquite different from others in the literature both in its purposeand approach. It is destined not for continuous multiple f0recognition but rather for projection of the ongoing spectrum tolearned pitch templates. The decomposition algorithm on the otherhand, does not compromise signal processing models for pitches andconsists of an algorithm for efficient decomposition of a spectrumusing known pitch structures and based on sparse non-negativeconstraints. After introducing the algorithm along with evaluations,a real-time implementation of the algorithm is provided for freedownload on the MaxMSP realtime programming environment. |
Mots-clés |
Multiple Pitch Detection / Realtime / Sparse Coding / Non-negative constraints |
Equipe |
Interactions musicales temps-réel |
Cote |
Cont06b |
Adresse de la version en ligne |
http://articles.ircam.fr/textes/Cont06b/index.pdf |
|