Research scientist PhD Candidate (CIFRE).

  • Full-time

Company Description

We are music and tech fans hailing from all over the globe, working to make Deezer the most personal music streaming service.

From data scientists to tech experts, artists & labels specialists to marketers, and even in-house music editors, our team is spreading the love for music to over 180 countries. Supporting local and international artists and bringing them closer to their fans is our mission - we believe music is about diversity, multiculturalism and togetherness.

Ready to join the team? We're all ears.

Job Description

PhD topic

The field of music information retrieval (MIR) has recently witnessed a breakthrough with the generalized use of deep learning architectures. These systems have considerably improved upon the state of the art in long-standing MIR tasks, such as chord estimation, instrument recognition, structural segmentation, or beat tracking. However, a major drawback of deep learning in its current form resides in the need for human supervision. In the case of music, collecting reliable annotations from polyphonic recordings is costly and unscalable.

In recent years, the need for annotations has been circumvented by the emerging field of self-supervised learning (SSL). SSL build annotation from the raw data itself and usually learn a low-dimensional representation of the data (in the so-called pretext task) that can then be used for solving a task of interest (downstream task).

The purpose of this PhD will be to explore the design of new pretext tasks leveraging pre-trained source separation models, in order to improve the state-of-the-art for several downstream tasks, especially in scarce annotation scenarios. Several downstream MIR tasks will be considered such as (but not limited to) downbeat tracking, chord detection, melody extraction, instrument detection, genre and mood recognition… 

Deezer Research

Deezer Research is a team of scientists and engineers working on Music Analysis, Information Retrieval, Machine Learning, and Recommendation. We do extensive research on challenging problems arising from dealing with huge collections of audio data and serving this to millions of diverse customers. During this PhD, you will work hand in hand with top-level scientists and engineers whose work has led to dozens of scientific publications in the past years.

 

Please send a resume and a cover letter specifying your motivation for doing a PhD in MIR/ML with your application.

Qualifications

What we are looking for:

  • Master's degree in Computer science / Applied mathematics or any quantitative field,  with a strong experience in Machine Learning or audio signal processing.
  • Fluent in spoken and written English.
  • Good programming skills, knowledge of Python and machine learning frameworks.
  • Knowledge of music theory is a plus.

Additional Information

Life @ Deezer Paris

  • Start-up environment and philosophy
  • Highly motivated and product-focused people ready to drive innovation
  • In-house Deezer Sessions with your favorite artists, gig tickets
  • Hackathons & meetups
  • Thursday drinks, summer and winter parties
  • A stocked kitchen with free drinks and snacks
  • Areas to relax and collaborate with beanbags, guitars and table football
  • An ‘at home’ vibe, with great outdoor spaces
  • Gym access at Deezer HQ, with lunch-time yoga and pilates classes
Privacy Policy