En poursuivant votre navigation sur ce site, vous acceptez le dépôt de cookies dans votre navigateur. (En savoir plus)
Portail > Offres > Offre UMR9912-PHIESL-001 - Post-doctorat H/F en modèles génératifs probabilistes appliqués à la musique

Post-doctoral fellow M/W in probabilistic generative models applied to music

This offer is available in the following languages:
Français - Anglais

Date Limite Candidature : lundi 15 mars 2021

Assurez-vous que votre profil candidat soit correctement renseigné avant de postuler. Les informations de votre profil complètent celles associées à chaque candidature. Afin d’augmenter votre visibilité sur notre Portail Emploi et ainsi permettre aux recruteurs de consulter votre profil candidat, vous avez la possibilité de déposer votre CV dans notre CVThèque en un clic !

Faites connaître cette offre !

General information

Reference : UMR9912-PHIESL-001
Workplace : PARIS 04
Date of publication : Monday, February 22, 2021
Type of Contract : FTC Scientist
Contract Period : 9 months
Expected date of employment : 1 April 2021
Proportion of work : Full time
Remuneration : between 2728 et 3000 € monthly gross depending on experience
Desired level of education : PhD
Experience required : 1 to 4 years


The post-doctoral fellow will develop new algorithms for learning multimodal embedding spaces, linking together symbolic, acoustic and perceptual representations of orchestral effects. To that end, the fellow will adapt zero-shot learning to musical content by developing architectures specifically tailored to the nature of audio signals
and symbolic scores. Therefore, specific transforms will be developed to incorporate multivariate time series struc- tures. The second part of the position aims to extend embedding spaces through variational learning with a specific information content measure, in order to discriminate the different dimensions of these spaces, and also allow for processes that generate music directly from the spaces. Finally, the fellow is expected to explore these regularities in musical data to study the underlying metric relationships and perform semantic inference on the data. This implies the creation of functional and usable interfaces for musical creativity, and testing these new softwares directly with composers.


The candidate will have to handle :
- Algorithms for learning multimodal embedding space
- Adapting zero-shot learning to musical content for both audio signals and symbolic scores.
- Assessing embedding spaces through variational learning
- Exploring metric relationships and semantic inference on musical data.
- Creating simple interfaces for these approaches
- OSC communication system for interfacing with Max / MSP and OpenMusic
- Creating packages for demonstration and final distribution
- Musical experiments and applications.


The candidate shall demonstrate
- Excellent knowledge of Bayesian inference and variational approximation. - Excellent knowledge of optimization and heuristic.
- Excellent Python skills and programming style.
- Good knowledge of probabilistic normalizing flows.
- Good knowledge of signal processing.
- High productivity, methodical and creative work
- Independence and good communication.
- Musical knowledge is a plus.

Work Context

Within the Research and Development department, the Musical Representation (http://repmus.ircam.fr) research team develops innovative algorithms for interfaces of musical composition, production, improvisation and analysis. The ACIDS group (acids.ircam.fr) will be the main host.

Constraints and risks


Additional Information


We talk about it on Twitter!