skip to content

Centre for Music and Science

 

Computational music cognition

The goal of computational music cognition research is to develop computer programs that simulate the psychological processes underpinning human musicality. This might for example include the perception of metre, the elicitation of expectation, or the evocation of beauty. Our approaches include psychoacoustic models grounded in signal processing, cognitive models grounded in symbol manipulation, and connectionist models grounded in deep learning. By developing and validating such models, we hope both to set music psychology on a rigorous mathematical foundation, and to create exciting software tools that can empower artists to create music in new ways. Current projects in this theme include: 

  • Acoustics and emotion
  • Consonance and dichotic listening
  • Consonance and timbre
  • Deep learning models of musical expectation
  • Pitch and categorical perception
  • Timing in organ performance
  • Timing in jazz ensembles
  • Tonality and emotion

Scaling up music psychology research with PsyNet

Our new software PsyNet enables music psychologists to run advanced behavioral experiments ranging from adaptive psychophysics to simulated cultural evolution. It provides many useful features for music experiments, including audio pre-screeners, stimulus generators, and online recording analysis. For more information see link.

Testing individual differences with psychTestR

Our psychTestR software has been adopted by music laboratories across the world as a tool for testing individual differences in music psychology studies. A broad range of music listening tests (e.g. melody discrimination, beat perception, tuning perception, rhythm perception) and self-report measures (e.g. Goldsmiths Musical Sophistication Index, Home Music Environment, Flow Experience, Music Engagement) have been implemented in psychTestR, as well as a large number of non-musical performance tests and psychosocial questionnaires. These form the basis of several ongoing large-scale music testing initiatives, such as the LongGold schools project and the Harvard Musical IQ project. For more information see link.

Score design for music reading: Cognitive and artistic perspectives

This project examines the cognitive processes that musicians use to read musical scores, and aims to propose new design principles to facilitate the reading of such scores. Our work so far suggests that simple but systematic and structured modifications to standard notation — such as the insertion of vertical blank spaces across the staff systems delimiting informational units — can lead to increased fluency and accuracy in sightreading. We are currently extending this work to broader ranges of expertise and musical repertoires; we are also teaching musicians how to use digital music processing systems to reconfigure their own scores in these optimised ways. For more information see link.

CMS Logo

Latest news

Huw Cheston starts internship at Spotify

10 June 2024

Good luck to Huw Cheston, current CMS PhD student, who starts an internship with Spotify today! Huw will be developing new machine-learning models for better understanding how different people contribute individual stylistic elements to musical recordings.

Xiaoyue Zhou to start funded PhD at University College London

13 May 2024

We are very happy to hear that Xiaoyue Zhou, summer project student at the CMS (2023), has been awarded a PhD studentship at University College London. Xiaoyue will be working with Prof. Maria Chait and Prof. Neil Burgess. Congratulations Xiaoyue!