Perceptually motivated deep learning approaches to creative sound synthesis
This project aims to facilitate creative sound synthesis with novel deep learning approaches, directed by interdisciplinary research into timbre perception and its semantic associations. Its aim is to develop a broader concept of timbre that generalises to electronic and synthesised sound, and to apply this understanding to develop new creative tools that allow for intuitive and perceptually congruent discovery, exploration and creation of sound.
C4DM theme affiliation:
Music Cognition, Sound Synthesis
Object-Oriented Programming (Undergraduate)
Major topics include the concepts of class, object, method, subclass, inheritance and their use in programming. The relevance of the object oriented style with respect to concrete software problems will be stressed both in lectures and labs. There will be two hours of lectures per week, and each student will have a weekly timetabled lab session. In addition, you will be expected to spend further time outside scheduled lab periods in the lab (or at home machines if they are available), and to read textbooks and review notes.
Neural audio synthesis; generative models; probabilistic modelling; timbre perception; psychoacoustics.