2014 10 23, Gerhard


On complication

During my studies in Vienna I composed my first computer generated piece based on a sound recording of two billard balls colliding (Der Zufall geht, 1986, stereo, 2'18"). I used only the first 256 samples (i.e. only a few milliseconds) of the collision sound as material for the whole piece. When played back over loudspeakers, this fragment was long enough to clearly identify the sound as a billard ball collision.


One day I discovered that the source of the click was not identifyable when the signal was played via headphones. It sounded just like a colorless, anyonymous and sharp click and I would not be able to say what its source was – the typical billard quality was completely lost. First I thought I had made a mistake and my program truncated the sound but when I listened to the same signal again via loudspeakers, the billlard quality was back.

The next audio contains the 256 samples of collision. Compare playing it back via loudspeaker and headphones.

The reverbation effect can also be simulated with the headphones by convolving the click with a binaural room impulse response. Compare the two audio clips using headphones. In the auralisation the click is played at a speaker left to the listening position.

It apears that the reflections of the signal in the listening space add a (random) structure to the source signal (essentially sustaining it) that allows our audition to reveil the source of the sound. There is actually no information added, as the reflections, which are essentially slighly modified copies of the original sound projected by the loudspeaker, can be considered redundant. But even in a dry room, the room response (reverberation) prolongs the signal substantially – at least by a factor of 10 and up to a 100 and more – giving the ear more time to get aware of the characteristics of the click.


One may argue that we know the sound of a billard ball only in its reverberated form and can therefore not recognize it if it is deprived of the room reverberation (i.e. if we played billiard in an anechoic chamber, it would not sound like billiard). This may be part of the phenomenon, but there is also a clear difference in pitch perception between the two reproduction situations. With the room reverb we can perceive the pitch (or rather the centre frequency of the burst) much clearer than without. The most characteristic part of the collision sound (the burst) consists of about 2.5 periods of a windowed sine wave with about 2000 Hz.


The reverberation repeats and thus sustains these 2.5 periods by the said factor, helping the ear to detect the pitch. The sustaining effect works even though the reflections of the burst are asynchronously (or statistically) superimposed in time. Although the sustained signal is essentially bandpass filtered noise with a centre frequency at about 2000Hz, this signal helps us to estimate the pitch much clearer than with the burst alone.

The reverberation of the burst, its temporal dispersion into discrete echos (lower order reflections), which after a short time (< 100 ms) form a dense reverberation decaying exponentially in amplitude, can be understood as a complication. There is no more information in the reverberated burst than in the dry one. But the relevant information (here the charateristic pitch of the billiard collision, which is determined by the material and size of the balls) can be perceived easier through the complication.

I wonder to what an extent such an approach can be generalized and used to explore certain features of structures through forms of complication rather than simplification or abstraction. Or do we have to understand the complication, which lets appear certain features better (and others worse) as a king of filtering and reduction – reverberation being a kind of convolution (i.e. filtering) in the end?

In the CoS project the practice of in-situ composition has been established. It consists in creating the work in the performance venue (and not in the studio) in order to be able to work with the interaction of the material with the venue, essentially the acoustics of the room – a particuar form of complication. Being able to choose, transform and arrange sound material under the acoustic conditions under which it will be experienced by the audience allows to create highly site-specific work, focusing on the interaction of the material, the composed processes and the site.

Applying this strategy to audification should be tested, both using the many speakers in the Ligeti hall as well as the icosahedron speaker at IEM, which has 20 individual speakers.