Enforcing harmonicity and smoothness in Bayesian non-negative matrix factorization applied to polyphonic music transcription Contraintes d’harmonicité et de régularité temporelle dans les approches bayésiennes de la factorisation en matrices coefficients positifs appliquées à la transcription de musique polyphonique
نویسندگان
چکیده
This article presents theoretical and experimental results about constrained non-negative matrix factorization (NMF) in a Bayesian framework. A model of superimposed Gaussian components including harmonicity is proposed, while temporal continuity is enforced through an inverse-Gamma Markov chain prior. We then exhibit a space-alternating generalized expectation-maximization (SAGE) algorithm to estimate the parameters. Computational time is reduced by initializing the system with an original variant of multiplicative harmonic NMF, which is described as well. The algorithm is then applied to perform polyphonic piano music transcription. It is compared to other state-of-the-art algorithms, especially NMF-based. Convergence issues are also discussed on a theoretical and experimental point of view. Bayesian NMF with harmonicity and temporal continuity constraints is shown to outperform other standard NMF-based transcription systems, providing a meaningful mid-level representation of the data. However, temporal smoothness has its drawbacks, as far as transients are concerned in particular, and can be detrimental to transcription performance when it is the only constraint used. Possible improvements of the temporal prior are discussed.
منابع مشابه
Décomposition arborescente et cohérence locale souple dans les CSP pondérés
Plusieurs approches récentes pour résoudre les modèles graphiques (réseaux Bayésiens avec contraintes) exploitent simultanément une décomposition du graphe et le maintien d’une propriété de cohérence locale. La décomposition de graphe exploite la structure du problème, offrant des bornes sur la complexité spatiale et temporelle, tandis que la propagation des contraintes dures conduit en pratiqu...
متن کاملNon-Negative Matrix Factorization Applied to Auditory Scenes Classification
This master's thesis is dedicated to the automatic classification of auditory scene using non-negative matrix factorization. A particular attention is paid to the performances achieved by the non-negative matrix factorization in sound sources detection. Our intuition was that a good classification could be achieve if we could efficiently detect the sources within auditory scenes. It appears on ...
متن کاملIncremental Multi-Source Recognition with Non-Negative Matrix Factorization
This master's thesis is dedicated to incremental multi-source recognition using non-negative matrix factorization. A particular attention is paid to providing a mathematical framework for sparse coding schemes in this context. The applications of non-negative matrix factorization problems to sound recognition are discussed to give the outlines, positions and contributions of the present work wi...
متن کاملفایل کامل مجلّه مطالعات زبان فرانسه دو فصلنامه علمی پژوهشی زبان فرانسه دانشکده زبانهای خارجی دانشگاه اصفهان
Tâ ÇÉÅ wx W|xâ Revue des Études de la Langue Française Revue semestrielle de la Faculté des Langues Étrangères de l'Université d'Ispahan Cinquième année, N° 8 Printemps-Eté 2013, ISSN 2008- 6571 ISSN électronique 2322-469X Cette revue est indexée dans: Ulrichsweb: global serials directory http://ulrichsweb.serialssolutions.com Doaj: Directory of Open Access Journals http://www.doaj.org ...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2009