1st MIP-Frontiers Workshop

April 30, 2020

General Information

The purpose of this first MIP-Frontiers Workshop was for the PhD students (ESRs) to give brief updates on the collaborations they had started during the “Sandbox” event that was held at QMUL, London, in February 2020, where they teamed up in groups of two or three to identify and work on some research question of common interest between their individual PhD projects. For the workshop, they were asked to prepare short (10 min) presentations on how their collaborative projects had progressed in the meantime, or why they turned out to be too ambitious or failed in other interesting ways.

The workshop was originally planned to take place at UPF, Barcelona, on May 9, right after the ICASSP 2020 conference. However, due to the Coronavirus / Covid-19 pandemic, it had to be turned into a purely virtual event. It was held via tele-conferencing (zoom). The presentations were recorded. The corresponding videos can be accessed in the workshop programme below, or at the dedicated videos page.

MIP-Frontiers is a multidisciplinary, transnational and cross-sectoral European training network for MIR researchers, funded by the EU’s Horizon 2020 program. More information can be found here.

Virtual Workshop Programme

Thursday, April 30, 2020, 10:00 - 12:45 CET

Welcome and brief project update

  • Welcome to the Workshop
    Gerhard Widmer, Johannes Kepler University of Linz
  • MIP-Frontiers in these Corona Times: A Short Status Update
    Simon Dixon, Queen Mary University of London

ESR group presentations

  • Conditioned-U-Net for Singing Voice Separation
    Giorgia Cantisani and Kilian Schulze-Forster
  • Real-Time Recitatives-to-Score Alignment
    Charles Brazier and Emir Demirel
  • A Deeper Dive into the Volume Sensitivity Issue of Singing Voice Detection
    Vinod Subramanian and Furkan Yesiler
  • Can Transient/Non-Transient Separation Improve Instrument Recognition?
    Carlos Lordelo, Alejandro Delgado and Antonio Ramires
  • Demystifying Differentiable Digital Signal Processing (DDSP)
    Ondřej Cífka and Javier Nistal
  • Automatic Identification of Jumps in Music Performances
    Luís Carvalho and Ruchit Agrawal
  • Cross-Dataset Evaluation of Auto-Tagging Models
    Karim M. Ibrahim and Philip Tovstogan