Music Information Computing Projects
AlgoRhythm - Developing a rhythm game for ameliorating sensory processing deficits of children.
We develop and evaluate a serious rhythm game for detecting and ameliorating sensory processing issues of children with autism spectrum disorder (ASD). Pilot studies indicate that therapeutic interventions based on musical rhythms can improve planning and execution strategies, but also improve childrens' social skills. This is a Dynamics-of-Youth project carried out in collaboration with the University Medical Center Utrecht and Leiden University, which brings together experts in music information computing, psychiatry, music therapy, game design, data science, learning technology, and neuroscience.
Find more information on this website or contact Anja Volk.
Cantostream - Computational analysis of tonal structures in early music
CANTOSTREAM is a corpus-based computational study of music composed between 1400 and 1750. During the 16th and 17th centuries a transition took place from modality, with its emphasis on melodic structures, to harmonic tonality governed by chord progressions. Our research focusses on the quantitative elements that characterise this transition.
Our main activities are:
- Measure human perception of musical boundaries
- Analyse the tonal patterns around these boundaries
- Build representative samples for a corpus study of 10.000 recordings
- Analyse music audio files with music processing techniques
If you are a UU-student and you want to collaborate, you can find more information about this project on our Konjoin project page. If you are from outside the UU, you can send a message to m.e.visscher@uu.nl or f.wiering@uu.nl.
Automatic Discovery of Patterns in Music
Musical pattern discovery algorithms have been researched for several decades in Music Information Retrieval. This project contributes to the comparisons between musical pattern discovery mechanisms by
- Developing digital interfaces for collecting human annotations of musical patterns
- Introducing new methods for the comparison between humanly and automatically discovered musical patterns
- Implementing Pattrans in Haskell for comparing musical pattern occurrences through musical transformations
- Comparing musical pattern discovery algorithms employing transformations between pattern occurrences.
This project is a collaboration between the Music Information Computing and Software Technology groups. For more information please contact Iris Ren or Anja Volk.
Bongo Beats - Tap with Me
Visually impaired children face many barriers in our society, and when it comes to social engagement, they engage less in cooperative play compared to sighted children. This project investigates music-based collaborative multiplayer computer games as a tool to encourage children to develop their social skills. Our game “Bongo Beats: Tap With Me” uses musical interactions to encourage inclusive play between visually impaired children and their sighted peers. Challenges for the game design include the quantity and quality of feedback which is necessary for the players, implementing adaptive difficulty to ensure the game is engaging for players of any skill level, and employing music information retrieval methods to extract beat information that aligns with perceptual qualities of the music. We collaborate with experts in the field of visual impairment, such as with the Bartimeus Foundation and Visio. For more information contact Anja Volk.
Computational Modelling of Melodies
Melody is fundamental and central in almost all music. In this line of research, we focus on computational modelling of melodic structures. One focus is on melodic similarity among traditional songs from various geographic regions. By computing similarities between various traditions, we can trace historic influences and discover the origins of particular melodies. We also work on melodic segmentation and on implied harmony of tonal melodies. We focus on various data sets. We heavily use the melody collection from the Dutch Song Database, hosted by the KNAW Meertens Institute, with which we have a long-standing collaboration. We also study melodies that are used for reading religious texts from Jewish, Christian and Islamic traditions. This work is part of a collaboration with the University of Bergen (Norway). One of the core questions is how the (improvised) melodies relate with geographic and religious identities. The Music Information Computing group contributes with computational methods to analyse audio recordings of readings in terms of scales and melodic patterns. In collaboration with Gregoriana Amsterdam, we study the relations between various Medieval European chant traditions. For more information contact Peter van Kranenburg.
Music Heritage as Linked Open Data
Building on the results of the H2020 Polifonia project [www.polifonia-project.eu], we focus on the accessibility of musical heritage data. By harnessing various techniques ranging from semantic web and ontology design to large language models, we build rich representations of historic music objects, including compositions and musical instruments, with a particular focus on pipe organs. By employing a disciplined co-design strategy, we explore ways to fit interfaces and data-presentation to the needs and knowledge of potential users, which allows both professionals and lay persons to interact with our rich cultural heritage. For more information contact Peter van Kranenburg.