When audio features reach machine learning - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2015

When audio features reach machine learning

Résumé

We review here our work on using machine learning in the Ecoute, Quaero and BeeMusic projects to perform automatic labelling into music-tags (genre, mood and instrumentation). We discuss the evolution over times of audio feature design ranging from manually designed audio features to automatically designed audio features. We also discuss the way we deal with scalability issues in these projects.
Fichier non déposé

Dates et versions

hal-01254057 , version 1 (11-01-2016)

Identifiants

  • HAL Id : hal-01254057 , version 1

Citer

Geoffroy Peeters, Frédéric Cornu, David Doukhan, Enrico Marchetto, Rémi Mignot, et al.. When audio features reach machine learning. International Conference on Machine Learning - Workshop on "Machine Learning for Music Discovery", International Machine Learning Society (IMLS), Jul 2015, Lille, France. ⟨hal-01254057⟩
115 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More