Melody Recognition with Learned Edit Distances

Amaury Habrard (1) , Jose-Manuel Inesta (2) , David Rizo (2) , Marc Sebban (3)

1 LIF - Laboratoire d'informatique Fondamentale de Marseille - UMR 6166
2 LSI - Departamento de Lenguajes y Sistemas Informáticos [Alicante]
3 LHC - Laboratoire Hubert Curien

In a music recognition task, the classification of a new melody is often achieved by looking for the closest piece in a set of already known prototypes. The definition of a relevant similarity measure becomes then a crucial point. So far, the edit distance approach with a-priori fixed operation costs has been one of the most used to accomplish the task. In this paper, the application of a probabilistic learning model to both string and tree edit distances is proposed and is compared to a genetic algorithm cost fitting approach. The results show that both learning models outperform fixed-costs systems, and that the probabilistic approach is able to describe consistently the underlying melodic similarity model.

Domaines: Apprentissage [cs.LG]

Amaury Habrard, Jose-Manuel Inesta, David Rizo, Marc Sebban. Melody Recognition with Learned Edit Distances. Structural, Syntactic, and Statistical Pattern Recognition, Joint IAPR International Workshops, SSPR 2008 and SPR 2008, Dec 2008, Orlando, United States. pp.86-96. ⟨hal-00322432⟩