Closing Comments
Background to Tao's design and implementation
Deficiencies in Tao's synthesis language
The computational expense of Tao's synthesis engine
Tao takes what has been referred to as the `brute force' approach to physical modelling and as such is not as efficient as some of the digital waveguide models developed by Julius Smith et al. However at the time of designing Tao I made a conscious decision to steer away from the obsession with real-time performance and look at what would be possible if I just concentrated on the conceptual structure of the system. First and foremost I wanted to design a system which made instruments which were tangible objects. As a musician I am accustomed to being able to experiment with sound in a direct, physical and intuitive manner and having a system which was capable of visualising the instruments was quite high on the list of priorities. At least if I couldn't actually get my hands on the instruments I could see them, which would in turn fuel my imagination for things to try out.Real instruments such as stringed or percussion instruments have a wonderful property that their 'user interface' is spatially distributed, and more importantly, doing things to the instrument at different spatial locations leads to markedly different timbral results.
In my DPhil thesis I argued that details which make a huge difference to the aesthetic appeal of the sounds produced from a physical model are often simply missed out in the name of real-time performance. I do not make any great claims about my model being so much more mathematically accurate than any others simply because it needs more processing power, but one thing I would say is that in the majority of cases musicians and composers who have heard the sounds which Tao is capable of producing have commented on their organic and inherent musical qualities, which cannot be a bad thing. Besides, with the exponential growth in computing power, the number crunching needed by Tao simply may not be an issue in the near future.
The calculations employed to animate the model are described in detail in my thesis. As they stand I am sure that improvements could be made in the name of efficiency and optimisation although they have already been optimised to an extent. There may be ways in which the efficiency could be radically improved without compromising the quality of the sounds produced and indeed I would be very interested to hear any ideas from individuals more mathematically skilled than myself. Tao is based upon mathematical skills I picked up at school level, so there should be room for improvement!
Closing Comments
Background to Tao's design and implementation
Deficiencies in Tao's synthesis language