New User Interfaces for Timbre
AimThe aim of this project is to investigate and address a problem in the user interfaces of electronic music synthesizers. While sound qualities such as pitch and loudness lend themselves relatively easily to user control, the timbre of sound is much less easily described or defined in any systematic way. Consequently, current user interfaces for controlling or modifying timbre in electronic synthesisers show a wide semantic gap between the task and system languages used. A key part of the objective is to design, implement, and evaluate a prototype of a new kind of user interface for controlling musical timbre that addresses this problem in a principled way.
The ProblemThe user interface of most synthesisers tends to be expressed in system (i.e. engineering) terminology, obliging the user to become fluent with the synthesis method employed. The problem of designing a more user-oriented user interface has previously been approached in a number of ways: using techniques drawn from artificial intelligence; knowledge based systems, and by employing metaphors derived from physical models of acoustical mechanisms. In these approaches, artificial intelligence and knowledge-based approaches to mapping user and system domain languages have generally focussed on automatic timbre classification strategies or acquisition of synthesis values by parameter extraction. These techniques can be helpful where a sample exists of the desired timbre, but they do not help much where the user is trying to specify a mentally imagined timbre. Similarly, physical models have been successful in providing the means of emulating ‘real’ instruments. However, an imagined sound for which the musician can find no physical analogue is one that cannot be easily realised using this approach. n this project, we are employing the notion of evolutionary design interaction to allow the user to act as the fitness function for a user interface driving a timbre generator that allows the rapid navigation of chosen timbre spaces.
(Click on any image to enlarge.)