SSM-NET: FEATURE LEARNING FOR MUSIC STRUCTURE ANALYSIS USING A SELF-SIMILARITY-MATRIX BASED LOSS - Télécom Paris Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

SSM-NET: FEATURE LEARNING FOR MUSIC STRUCTURE ANALYSIS USING A SELF-SIMILARITY-MATRIX BASED LOSS

Résumé

In this paper, we propose a new paradigm to learn audio features for Music Structure Analysis (MSA). We train a deep encoder to learn features such that the Self-Similarity-Matrix (SSM) resulting from those approximates a ground-truth SSM. This is done by minimizing a loss between both SSMs. Since this loss is differentiable w.r.t. its input features we can train the encoder in a straightforward way. We successfully demonstrate the use of this training paradigm using the Area Under the Curve ROC (AUC) on the RWC-Pop dataset.
Fichier principal
Vignette du fichier
ISMIR2022_lbd_5.pdf (2.07 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03860497 , version 1 (07-07-2023)

Identifiants

  • HAL Id : hal-03860497 , version 1

Citer

Geoffroy Peeters, Florian Angulo. SSM-NET: FEATURE LEARNING FOR MUSIC STRUCTURE ANALYSIS USING A SELF-SIMILARITY-MATRIX BASED LOSS. Late-Breaking/Demo Session of ISMIR (International Society for Music Infor- mation Retrieval), Dec 2022, Bengalore, India. ⟨hal-03860497⟩
11 Consultations
50 Téléchargements

Partager

Gmail Facebook X LinkedIn More