Permalink to this page
Computational Modelling of Music Perception and Cognition
We are developing and testing models designed to explain, from first principles, our perception and cognition of music. Projects we are currently undertaking include:
• Algorithms to Discover Musical Patterns—Our research involves putting specific aspects of music perception, such as the discovery of certain types of musical patterns, on an algorithmic footing. The resulting algorithms are evaluated by comparison with human experts/listeners performing analogous tasks. We use disparities between algorithm output and human response to shed light on the more nuanced strategies behind human perception of music, thence to improve our models (Tom Collins, Robin Laney).
• Computational Modelling of Tonality Perception—A psychoacoustic model designed to explain the feelings of expectation and resolution that are induced by various chord progressions and scales. The model is currently being developed and experimentally tested by comparing the model’s predictions with those given by humans (Andrew Milne, Robin Laney, David Sharp).
• Melodic emotions: Insight and Prediction—Examining the role of melody in conveying emotions and whether there are characteristic melodic features that can be used to predict a melody’s emotional impact on listeners (Pauline Mouawad, Robin Laney, Chris Dobbyn).
• Testing Computational Models of Rhythm Perception using Polyrhythms—Rhythm perception models (including Ed Large’s non-linear dynamics model) are tested with rhythms containing two different pulses—such as three against four rhythms found in African-derived music (Vassilis Angelis, Simon Holland, Martin Clayton).