Modelling long-term dependencies in time series has proved very difficult to achieve with traditional machine-learning methods. This problem occurs when considering music data. In this paper, we introduce predictive models for melodies. We decompose melodic modelling into two subtasks. We first propose a rhythm model based on the distributions of distances between subsequences. Then, we define a generative model for melodies given chords and rhythms based on modelling sequences of Narmour features. The rhythm model consistently outperforms a standard hidden markov model (HMM) in terms of conditional prediction accuracy on two different music databases. Using a similar evaluation procedure, the proposed melodic model consistently outperforms an input/output HMM. Furthermore, these models are able to generate realistic melodies given appropriate musical contexts.
74
Views
3
CrossRef citations
Altmetric
be0ef6915d1b2200a248b7195d01ef22
research article
Predictive models for music
Jean-François Paiement Idiap Research Institute, Centre du Parc , Rue Marconi 19, Case Postale 592, CH-1920 , Martigny , Switzerland Google , 1600 Amphitheatre Pkwy, Mountain View , CA , 94043 , USA Correspondencepaiement@gmail.com, Yves Grandvalet Idiap Research Institute, Centre du Parc , Rue Marconi 19, Case Postale 592, CH-1920 , Martigny , Switzerland Heudiasyc , CNRS/Université de Technologie de Compiègne, Centre de Royallieu , B.P. 20529, 60205 , Compiègne , France & Samy Bengio Google , 1600 Amphitheatre Pkwy, Mountain View , CA , 94043 , USA
Page 253-272
Published online: 18 Nov 2010
Log in to Taylor & Francis Online
Or purchase it *
-
Add to cart
Article Purchase 24 hours access for EUR 39,00
-
Add to cart
Issue Purchase 30 days access for EUR 418,00
* Local tax will be added as applicable
People also read
research article
Exploiting functional relationships in musical composition