xCoAx 2020 8th Conference on Computation, Communication, Aesthetics & X 8–10 July, Graz online

Matters 5

Keywords: Acousmatic, Granular Synthesis, Multichannel Audio, Algorithmic Composition.

Sound and Form

Granular synthesis has been in the focus of my compositional work for many years – however, there are always variants, that open new possibilities, at least for the own perspective. In Matters 5 I decided for a relatively simple way of buffer granulation with SuperCollider patterns. While experimenting with synchronous and asynchronous granular sequences I ended up with a rhythmical structure of phrases that consist of percussive events, short synchronous sequences and rests. The latter follow a fixed exponential random distribution and the percussive events stem from overlapped grains. This pattern produces a kind of irregular regularity, a non-beat swing, that, with its specific parameter setting, I appreciated very much and which gave me the room to work with strange reverb effects. The rhythmical structure of the granulation is confronted with – respectively overlapped by – LFO-controlled movements of the buffer read head and further LFO controls of various reverb parameters, which produce meta waveforms and further near-rests.

As the rhythmical foundation seemed to be sustainable for different sound sources I decided for two different ones that lead to interestingly contrasting results. In the first part a scratch sound is taken and, due to its noisy nature, perceivable pitch, though present, doesn't play an important role. The grains' playback rates are randomly changing every few moments so that the overall harmonic character is atonal. Decreasing dampen parameters of the reverb are causing accumulations of broad clusters, further blurring any unique perception of pitch in a section of turmoil before one of relaxation. In the second part a short glissando gesture of the double bass, moving around a central tone, is taken as a base for combinatorically established harmonic variations in just intervals (3/4, 4/3, 7/8, 8/7). Its character is much more quiet, but a second bow form within the piece is built up. It results from a gradual shortening of grains, which changes the role of the double-dealing reverb to that of metal-like resonator. Single intermediate sequences of longer grains cause reminiscences of the glissando gesture.


The 24 speakers are treated as three rings of 8 speakers, which allows an adaption to different setups from 8 speakers onwards (see the chapter about routing convention). The spatial movements come from two different and independent algorithms: The scattering of grains is organised as a loop of weighted random choices – for percussive pulses and short sequences two or three L/R-pairs in each ring are chosen. The probabilities weaken the weight of rear and rear-side speakers. The continuously changing dampen and mix parameters of the reverb units (low frequency sine waves of close frequencies) are additionally causing spatial movements, which are noticeably accelerating in the first part of the piece. In the end two SuperCollider programs were rendered to two 24 channel audio files (26 including the stereo version). Besides a single crossfade in the middle of the piece and applying a low shelve filter no post-processing was done in the digital audio workstation (Reaper), see Fig.1.

Fig. 1: Matters 5, Reaper session, overview (12:46)

Fig.2 shows a turbulent section in the first part. On the left side the combinatorial distribution of percussive events in space can be observed, the middle shows the slightly decorrelated amplitude oscillation of the reverb, which becomes slower on the right side. The different weights of channel pairs can also be seen – in every group of 8 especially the rear channels 5 and 6 are of lower amplitude. The timescale of the snapshot inhibts a view into the granular structure and its separate spatial characteristics, which overlap the spatial effects of the reverb.

Fig. 2: Matters 5, Reaper session, ca. 2:32 – 2:54

Routing in Various Multichannel Setups

The mono sound files should be used according to the following rules:

  1. The basic 8 ring (soundfiles with suffixes A1, A2, ...) must be there and placed in the intended order, preferred in the low level.
  2. The optional 8 rings (soundfiles with suffixes B1, B2, ..., C1, C2, ...) can be placed at higher levels or at the same level as the A ring. There's no obligation to keep the B-ring below the C-ring, their roles can be switched. If levels have different numbers of speakers, then the groups A, B, C can be distributed to different levels as long as all other rules are fulfilled.
  3. The L/R-pairs of the optional rings (1/2, 8/3, 7/4, 6/5) must remain complete (e.g. 1 must be used with 2) and placed at left and right side of the room, mirrored at the axe.
  4. Pairs of optional rings might however be omitted, as long as the L/R-pair 1/2 is used and the order of the L/R-pairs from front to back is kept, e.g. (1/2, 8/3, 7/4) or (1/2, 6/5).
  5. The L/R-pairs 1/2 and 8/3 must not be placed in the back half of the room whereas the L/R-pairs 7/4 and 6/5 must not be placed in the front half of the room.

Example Setups

Fig. 3: full 24 speaker setup with 3 rings, for a 16 speaker setup with two rings the C-ring as well as the B-ring can be omitted.
Fig. 4: setup with incomplete B- and C-rings, both incomplete rings could be omitted as a whole.

Suggested Routing in the MUMUTH

From the speaker setups described in the call, "hemiaux" would be suited best. It can roughly be understood as a setup consisting of 3 rings of 10, 8 and 6 speakers, nearly mirrored along the room axe. It’s almost a perfect layering of 3 x 8 rings, as only adaption – in accordance with the described setup rules – the two rear sources of the C-ring would be routed to the additional rear speakers of the low-level ring.

Fig. 5: Matters 5, suggested channel routing for the MUMUTH.

Media Assets

Binaural version mp3:

Binaural version wav:

Join the conversation