Master 2018 2019
Stages de la spécialité SAR
Investigating statistical models of music perception with artificial neural networks and brain data

Site : Laboratoire des systemes perceptifs, equipe audition
Lieu : Ecole Normale Superieure, LSP, 29 rue d'Ulm
Encadrant : Shihab Shamma, Giovanni Di Liberto
Dates :du 18/02/2019 au 31/07/2019
Rémunération :554.40 Net/Mois
Mots-clés : Parcours ATIAM : Informatique musicale

Cliquer ici pour vous authentifier


In order to process music, humans learn its melodic structure and make continuous predictions of upcoming events. Recent work proposed that a statistical model based on hidden markov chains approximates this brain process (Pearce, 2005 ; Pearce, 2018). The internship aims to investigate whether state-of-the-art models for music generation, such as recurrent neural networks (Hadjeres, 2017), can provide us with a better description of this brain mechanism. In the first instance, we will derive models uniquely from music data and we will evaluate their physiological validity by using neural data recorded with electroencephalography, which was shown to reflect the processing of the statistical structures of auditory stimuli (Broderick et al., 2018). Next, neural data will be used to fine-tune the parameters the neural networks directly in a multimodal fashion, in order to derive an estimate that optimally fits the brain responses to music.


M. Pearce, “The Construction and Evaluation of Statistical Models of Melodic Structure in Music Perception and Composition,” PhD thesis, School of Informatics, City University, London, 2005.

M. Pearce, “Statistical learning and probabilistic prediction in music cognition : mechanisms of stylistic enculturation,” Annals of the New York Academy of Sciences, 2018.

Gaëtan Hadjeres, François Pachet, Frank Nielsen, "DeepBach : a Steerable Model for Bach Chorales Generation", arXiv, 2017

M Broderick, AJ Anderson, GM Di Liberto, MJ Crosse, EC Lalor, "Electrophysiological correlates of semantic dissimilarity reflect the comprehension of natural, narrative speech", Current Biology 28 (5), 803-809.