[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[AUDITORY] Fully funded PhD Fellowship in Machine Learning and Information Retrieval for Music-Related Movement

A fully funded PhD position is available in the Robotics and Intelligent Systems group at the Department of Informatics, University of Oslo, Norway. The main area of focus within the position is automated classification of motion capture data of dance recordings, including feature extraction and selection, and matching against semantic descriptions of the data.

The appointment is for a period of 3 years. There might be a possibility to extend to 4 years depending on the qualifications of the recruited candidate, and the department’s need for teaching and lab assistants. The recruited candidate will be working within the fourMs group and our recently awarded centre of excellence, RITMO.

Project description:

The field of Music Information Retrieval (MIR) has developed advanced strategies for analysing music as an auditory phenomenon. Strategies involve feature extraction and selection based on physical properties of the sound signal and also models of human perception. Another important aspect of music, which to a lesser degree has been subject to MIR research, is movement: Music starts with sound-producing movement, and often also results in movement in the form of dance.

Movement may be quantified precisely using motion capture technology. But how are quantitative representations of movement related to semantic descriptions of the same movement? Can a computer be trained to classify dance styles and dance genres? And is it possible for a computer or a robot to imitate human dance movement from audio?

In the announced project, the recruited PhD candidate will research machine learning techniques for full-body motion capture data. The work involves contributing to data collection using state-of-the-art motion capture technology from Qualisys. Further the candidate will use the collected motion capture data for various machine learning tasks:
  • Automated generic post-processing techniques for motion capture data (automatic marker recognition, gap-filling and marker swapping)
  • Classification of dance data based on semantic descriptions (such as dance genre/style, gender, expressivity, etc.)
  • Explore deep learning techniques for automated synthesis of dance movement.

Requirements and qualifications:

Applicants must have a Master’s degree in a relevant field such as computer science, machine learning, biokinematics or musicology/music information retrieval. A solid background in computer science and machine learning is required, as well as good analytical and programming skills. Experience with Matlab and MIRtoolbox/MoCap toolbox is preferable. Further, competence in several of the following fields is desired and will be considered an advantage when candidates are ranked: data analysis, digital signal processing, motion capture technology, and music.

Pay grade, depending on qualifications and seniority: NOK 436 900 - 490 900 per year 
(approx.: $56,300 - 63,300 / €47,000 - 52,800)

Full announcement:

Closing date for applications:  31th October, 2017

Contact for more information:  Assoc. Prof. Kristian Nymoen  E-mail:  krisny@xxxxxxxxxx


Kristian Nymoen
Associate Professor
Dept. of Musicology, and Dept of Informatics
University of Oslo
Office phone: (+47) 22841693
Mobile phone: (+47) 99708138