It's nice that we are again "with ourselves" in this answering procedure, again parallel everyone in their space, before we will look again at the other answers.


{kind: note}

From Ji:

From Daniele:

From David:

 

interaction

People pass through the stairs, either going up or down, or sometimes standing still. Do you see any different influences of their movements especially on the segmentations within your musical outcome? In other words, can (a) segmentation(s) become a flexible or an organic entity by their interactions?

 

{kind: quote, persons: JYK}

counting

What is the maximum number of segments can you drag out of the site? Can the way you count them be differentiated? Can you find a way to do it either intuitively, or using an existing counting technique? And what way is interesting the most? Can this counting method itself be organic?

 

{kind: quote, persons: JYK}

bridge(s)

Can (a) bridge(s) be included in your segmentation networks? If so, do they have a direction? :are they(is the one) supposed to be one directional, two, or more? If there are more than one bridge, do they look identical or not? Can they have a function as to make the whole network extended? Do they have the same characters or different from each other? Would they be recognizable, e.g. having a certain duration and standing as a segment?

 

{kind: quote, persons: JYK}

edges

So the terms I highlighted {new aesthetic results | a single entity | an amorphous zone | installation's edges are not always perceptible | absence of a recognisable boundary} sort of grasped my attention, and I think it could be interesting for us to include a similar reflection in our project. I think we already made a move in that direction by deciding to integrate the loudspeakers in the existing staircase architecture, but I would like to shift the focus a bit towards the acoustic situation.

  • How can we create a situation where the environment and our work are perceived as a single entity, rather than as two separate units (background / piece)?




  • What would it mean for our installation to have (acoustic) "edges" that are not always perceptible?




  • Is it possible, and do you find it meaningful, to think about our installation as an entity with no recognisable boundaries? What would that mean in concrete?




  • Amorphous zone: how could this concept be tranlsated into the sound domain?




  • By thinking in this terms - trying to establish the highest integration possible between the work and the space - would you expect to achieve some form of aesthetic result that is different from the sum of the two?

 

{kind: quote, persons: POZ}

affection

I find fascinating to work with synthesis processes that can generate sound forms out of implemented relationships. In my practice this is often achieved through bottom-up processes, in which I consider the modelling of the relationships among the elements that compose the system the crucial point of the work. Generally I use very few elements, and spend much more time and effort in designing and calibrating the joints among them. This also implies thinking how an element could affect the others and, viceversa, how that element could int turn be affected. There is a point where the design of these affection scheme takes precedence and occupies an aesthetic context in itself.

Basically, I'm trying to understand if it makes sense to introduce some of these ideas in our situation.

  • How could our 4 instances relate to one another, besides the acoustic relationship that will anyway happen simply by sharing the same physical space?




  • Which kind of data do we need to exchange, and what type of "receptors" shall we integrate in our systems in order to realise a form of mutual affection that goes beyond simple communication (OSC message->"hey, I'm playing now") - something that resembles more a structural coupling and that somehow occupies an aesthetic context in itself?





  • What's the difference between affection, influence and communication?

 

{kind: quote, persons: POZ}

presence / absence

To occupy: to fill, exist in, or use a place or period of time.

How can we think of the relation between the space we work in and the sound that will occupy it?




When do we perceive sound as "occupant"?




Can we distinguish at all between a presence or absence of sound?




If yes, what movements do articulate the transitions between the two?




What's the relationship between the space of a sound (its spectrum) and the sound in space?




And what the one between the temporal dimension of a sound and its perceived spatiality (presence/absence/movement)?

 

{kind: quote, persons: POZ,}

synchronisation / interaction

Which are the regions of effective interaction {with the existing patterns / rhythms of the space} ? How to find them?

 

{kind: quote, persons: DP}

 

being part / deviations

Can these deviations {our intervention in sound that modifies the existing soundscape} be somehow subject to composition? That is, how could it be possible to "compose" which "form" the deviations we will provoke have and how big these are?

 

 

 

 

Should we let the complete system of all installations and the space evolve autonomously to an new equilibrium, or highlight exactly this movement of deviation or of evolution enacted by the system?

 

{kind: quote, persons: DP,}

network / meshwork

Which strategies can we devise, that go beyond simple (first order) rules like "if the others play loud (or above a certain threshold), this process will attenuate its output"?

 

 

 

 

Is it viable to just exchange some channels of sound? It does not have to be the same sound the works are projecting into the space: maybe some intermediate stage of the sound synthesis?

 

{kind: quote, persons: DP}

parallel / scale invariance

May we think of a "parallelism" that transgresses different temporal scales? Similar processes that operate on an with (and through) different scales of time?

 

{kind: quote, persons: DP}

HHR: Responses Round 1

I was perplexed to see segmentation appearing so clearly as the operation implied in the question. More so, as I was also including segmentation algorithms in the original survey (e.g. image segmentation). Indeed I was thinking of clustering evolving spectra to obtain "gestalts". But after that I had put aside the thought about "segmentation operations" as the foundation of the piece. I like this idea very much, even though I do not know yet how this operationalisation would look like.

 

{function: response, keywords: [segmentation]}

I read "drag out" as a sort of extraction? I like the idea of counting as a fundamental algorithmic operation. As a kid, when we went on vacation by car, I used to play a game of counting specific objects, say other cars on the highway of a particular colour… Or counting the number of walking steps on a particular way through the city.

 

{function: response,  keywords: [counting, extraction]}

This combination segmentation—network I found again perplexing and compelling. So first we have the discreteness, then we have the relations, but now thought of as movement (a bridge leading from one to the other). When I think of network, I almost inevitably think of a graph and ways to traverse the graph. Rational numbers, breaking numbers apart, putting something between the atoms. The devil's staircase. We could also think of bridges as interpolation. Even plural: networks.

 

{function: response, keywords: [bridges, segmentation, network, discreteness, relationships, movement, interpolation]}

The blurring of edges reminds me of Morton Feldman's admiration of the paintings of Mark Rothko, which also had to do with the way the edges were faded. In terms of the architecture, I would first of all think of the sound skin—the Max Neuhaus installation, which also does this blending in with the environment (it might be a frame of reference for us). Another angle would be Robin Minard's distinction between articulating or conditioning an atmosphere (the former would probably correspond to the idea of the "single entity").

For me this question of blurring is related to the distinction (or collapse) of foreground/background—so we meet here directly with your question of background/piece.

On Sunday, we were walking up a small hill. The pattern of far away church bells was entirely blurred, so that you heard an only subtly modulated overall timbre of the mixed bells. Not unlike the Neuhaus piece (although admittedly even more beautiful because of the huge distance).

 

{function: response, persons: [Mark Rothko, Morton Feldman, Max Neuhaus, Robin Minard], keywords: [edges, architecture, articulation, conditioning, foreground, background]}

The relationships as joints make me think of Ji's bridges.

 

Do you actually want to use the term affection, or the noun for the verb to-affect, which would probably be influence? Since later down you enumerate both affection, and influence, is it not a stretch to apply affection to machine couplings?

 

The non-trivial ways of relating (the trivial way is what you already identify as the sharing of a reverberant space) are, in my opinion, the one you mention—the communication channel—and the process that we are undergoing right now (the simultaneous work on the layers, during which we share thoughts, concepts, strategies, code, …). The former should arise from the latter.

 

c

So I'm coming back after a gap of a few days. When I was in the workshop in Lisbon, Dieter Mersch—with whose view on computational art I have severe problems, but otherwise had some interesting exchange—asked after my presentation why, when we think about computation and algorithms, we come always to connected diagrams and graphs, and so if there was any thought on emptiness, a world of disconnected isolated things, and what about the horror vacui that according to him was at the core of all art.

 

I remembered that in the construction of Almat, I was thinking about this problem of "growth" and "filling" of algorithms and generative processes, and how we could come to think of shrinkage. Throw a positron and an electron together, and they produce annihilation (but that is not void, right?)

 

About occupation. Jean-François Lyotard wrote an essay "Oikos", which deals with the problem of systems and their boundaries, the distinction inside/outside, and so on, and there he mentions the character of the "ignored guest":

 

My oikeion is an otherness that is not an Umwelt at all, but this otherness in the core of the apparatus. We have to imagine an apparatus inhabited by a sort of guest, not a ghost, but an ignored guest who produces some trouble, and people look to the outside in order to find out the external cause of the trouble. But probably the cause is not outside, that is my idea. So we can call it entropy, but probably the more interesting thing is to try to touch it, not approach it, because it is not an object available for a cognitive touch.

(by the way, here is another example of "apparatus" which does not require you to think it just in Foucault'ian terms)



{function: response, persons: Mersch,  keywords: [emptiness, shrinkage, boundaries, inside, outside, occupation, apparatus]}

I am not sure that the extraction of the patterns arises from "acoustical interaction". Acoustics are always involved, but I do not think that the patterns you are talking about, such as steps or the frequency with which doors are opened and closed, have a tight coupling to the acoustic properties of the space.


{function: response, keywords: [patterns, interaction, coupling, space, acoustics]}

I think Ji's idea of segmentation answers this question with a clear "yes", it is possible to compose the form of interventions. There are many other ways we can think of this, for example severing the temporal from the timbral behaviour.

 

{function: response,  keywords: [segmentation, form , composition, behaviour]}

I do not think that we need to go beyond such rules as a starting point. It does not appear to me that such rule is a "first order" rule—in the sense that it forms the lowest in a hierarchy, and this is less useful than any other rule. We agree that even the simplest rules can produce very complex behaviour, so I would tend towards beginning even with rather simple rules. (In other responses, I want to highlight that I think not so much about rules that one has to obey but a bidirectional communication).

 

{function: response, keywords: [rules, communication]}

Yes, I very much like this idea of a fractal form, because it just describes a relation within itself, without pointing out (yet) what the elemental operation is.

 

{function: response, keywords: [fractal, form]}

preliminary thoughts 15-dec-2019

preliminary thoughts 10-dec-2019

Rayuela

I try to verbalise these rather quickly (whatever comes to my mind without much filtering), before then returning to select questions in depth.


{kind: note}

I would like to avoid a solipsistic piece, so hopefully we will find a way to formulate meaningful responses to the movement of people in the space; where "meaningful" probably refers to a way that people can experience the responsiveness. This could be a utopian attractor, as I am not sure that for example relative motion up/down the stairs can be extracted from the four microphone signals (perhaps it can). In any case, I would affirm that whatever the form of segmentation, it should be flexible and organic with respect to the environment.

 

{function: response, keywords: [motion, visitors, interaction, microphone, form, organic, environment]}

Counting is registration first of all. Geiger counter. Obtaining the frequency of events (discreteness). Counting each stone or counting the joints. Counting the pavement stones based on the ratio of the stone size and your gait. The staircase: taking each step versus taking every two steps (hectic). Counting backwards I count you in. (Thus: counting towards an event - a different quality of counting than the pure movement of numbers). Gaming pieces.

 

{function: response, keywords: [counting]}

Rather than a graph, perhaps it's better to think of the space of segments as a topology. Naturally, as a "space", you will always be able to get from one segment to the other, perhaps through a number of intermediaries. I have never worked with Petri nets. Bridges are called arcs here, vertices are called places (so quite literally the net is a topology). There is a direct notion of parallelism (concurrency) in petri nets. Similarly, in P-systems, segments would be membrane structures, and bridges could be the rules for objects to travel inwards or outwards, or to dissolve a membrane (a collapsing bridge).

 

{function: response, keywords: [bridges, segmentation, network, topology, space]}

The most obvious margins of perception are of course the audible frequency range (excess going into ultrasound or infrasound), and loudness (with the dynamics of masking against the existing environment). But it could also be imitation, "blending in", camouflage, becoming transparent. But also active processes on our sound material, such as erasure and blurring. If something recedes into the background, it moves away from the layer of focus. There are also the curious games of attention: The ticking of the wall clock that disappears from our consciousness, the aural blindspots.

The idea of edges could also literally translate into the groupings of eight speakers.

 

I would argue that if there are no recognisable boundaries, there is no object. Objects are distinguished as boundary crossing. What is interesting is not the blindspot, but the moment when you recognise that something has disappeared from our perceivability. I once made a simple sound installation in the entrance of the ORF centre in Graz; you would hear field recordings from either inside or outside the building, which would naturally blend with the space as "probable". Then in intervals, the sound would just switch off, thereby marking the boundary. I think you need a game of hide and seek. Otherwise (this is of course also valid) you are simply altering the atmosphere and leave it to the believe of people what the nature of the intervention is. I wonder if this is not just sound design then?

 

{function: response, keywords: [margins, frequency, loudness, layers, focus, attention, boundaries, object]}

If you think from the relationality, the communication would be based a) on mutual agreements and b) individuated for each pair (poz ⟷ jyk, poz ⟷ dp, poz ⟷ hhr, jyk ⟷ dp, jyk ⟷ hhr, dp ⟷ hhr). Perhaps not even all of these six connections must be realised.

 

It is an open question what "simple" means, as well as whether simple is desirable or worth avoiding. I would tend to a small vocabulary.

 

{function: response, keywords: [communication]}

 

 

The question of absence/presence seems related to the edges, whether they fade, whether we perceive them. We can certainly introduce and outroduce a sound in a seemingless way, so there is only a change of presence/absence when we actively put that as a question to our observation. But absence/presence seems orthogonal to occupying/transparent. It is not a zero sum game. You can create a sound that does not "take away" space from other sounds and element, but can coexist with the rest. Occupation means occupation of attention. Occupation has some negative connotations, but we can view it as a neutral term or description. Something can take away our attention; that's a decision we make or make not.

 

Especially when our sounds depend on the sounds people in the space produce, "occupation" would thus not fall back just on our doing, but it is also an act of the visitors themselves. There are two possible reactions; either people would like to emphasise that occupation—stimulus response, acoustical selfies, a childish joy of effectuation. Or, in my experience less likely or more difficult to produce, people would start to behave more carefully to avoid creating sounds in the space.

 


{function: response, keywords: [presence, absence, edges, observation, attention, visitors, interaction]}

The acoustics of the space are more or less givens, even if slightly modulated by the presence of people or the coupling to neighbouring rooms as doors are opened. I think the ensemble of these is captured by the atmosphere or soundscape of the space, of which the acoustic shaping is then an integral part.

 

{function: response, keywords: [patterns, interaction, coupling, space, acoustics, soundscape]}

One process I find very interesting is based on this, by creating a "minimum phase" version of a chunk of sound, which preserves mostly the timbral qualities, although it is clear that time and frequency are related, so often there are remnants of the temporal behaviour as well, still resembling voice for example if the chunk is spoken language.

 

{function: response, keywords: [minimum phase]}

This seems to relate what I wrote somewhere above in response to Daniele's question on communication. That one layer signalises that "it plays loud" is a signal that each layer can interpret in a way useful to that layer. If that signal implies a rule on other layers, then this is not a form of communication but a command. I think here is the boundary of autonomity. At some point we will have to negotiate what to do if layers are "too loud" compared to other layers. But it should be the responsability of the respective "too loud" layers. Reponse-ability in the Donna Haraway sense.

 

So perhaps instead of a signal from layer A that "I play loud", there is rather a need for a layer B to signal "I think, you others are playing very loud, please respond to this".

 

{function: response, keywords: [autonomy, communication, control, rules]}

I would even interpret "scale" very loosely, in order not to constrain the idea of self-similar form to an overall heterogenous data model.

 

{function: response, keywords: [fractal, scale, form]}

(see next step for examples using minimum phase transformation)

A very early module in my "rendering" software FScape (so early 2000s) that I have always found very effective, although (or because?) it is so simple, is "Step Back". It tries to chunk a sound file into segments by looking at local maxima in decorrelation (as far as I remember), so finding points of change. You give it a minimum / maximum time range, within which to locate each such point of change. The segments are then usually just reassembled in time-reversed order, which unravels the macroscopic form, but retains the meso-level organisation and thus a degree of "organicity".

 

{function: response, keywords: [FScape, sound, module, chunk, segmentation, form, organicity]}

Tally marks. The age of people (the size of their feet, the speed of their steps). Counting into categories. Rhythm. Non-metric beats. Modulus and overflow. Greatest common divisor, least common multiple. Transactions (atomicity).

 

{function: response, keywords: [counting]}

In many graph systems, the edges seem somehow inferior to the vertices in their "substance". They may hold a label or "cost", but they are only there to connect vertices, not in their own right. That is somehow related to Ingold's critique of the network metaphor as opposed to meshworks which begin with relationships.

 

I would like to think of segmentation-networks as something entirely different.

 

{function: response, keywords: [bridges, segmentation, network, edges]}

I think a caveat here is that we are in a dedicated art space. The play of amorphous camouflage does not necessarily work here. There is a reason the Neuhaus sound skin evaporates outwards into the city soundscape, into an otherwise not intentionally aesthetisised space. What happened when Fluxus was put into a museum? Did that work out? We like to think there should be no boundary between art and life, but is that not late-modernist?

 

What does integration mean? One resolves within the other. It somehow stands contrary to the segmentary for me. I don't see as a problem that a sound intervention reconfigures the space, it does not need to become one with the space. What would be the motivation for this "monism"? One does not have to cover the other, they can intersect, they can be distributed, parallel.

 

Of course, the sound could elaborate the space. It could make people aware of their movement in the staircase, of the materiality of the concrete, of the lighting, of the resonances and reverberation. I would perhaps call this alignment instead of integration (it does not dissolve).

 

What are the edges between the four layers?

 

{function: response, persons: Max Neuhaus, keywords: [camouflage, integration, intersection, parallel, distributed, space, layers]}

Niklas Luhmann describes autopoetic systems as characterised by loose coupling. If we assume a certain autopoetic quality of each layer, they would be driven by a communication which Luhmann describes as "continual coupling of self-reference and hetero-reference".

 

Communication can be understood from the perspective of a "receiver" making sense (Elena Esposito). The term itself of course has been problematised based on metaphors commonly associated with it (Klaus Krippendorff). Nevertheless, it's important to distinguish technical communication from "real" communication. If one system tells another "I'm playing", that is strictly speaking not yet a form of communication, because it does not include the interpretation of that message on the receiving side. I think, ideally the communication system we establish has to do with mutuality, there needs to be a related bidirectional flow.

 

In what way a layer affects the other first rests with the "potentially affected" layer, and ultimately with an observer discerning that affectedness. I think what is relevant for us in the first place, is that there is mutual exchange of information, which undergoes a twofold selection process. Selection by the layer that selects data for transmission, and selection by the layer that selects how to deal with that received data.

{function: response, persons: [Niklas Luhmann, Elena Esposito, Klaus Krippendorff], keywords: [autopoiesis, communication, layers]}

I feel the last question is quite broad; I would not be able to give a general relatgion between the temporality of a sound and its spatiality. Therefore, I want to move to the relation of a sound's space or spatiality and the sound being in the space.

 

First of all, and this seems to be corrected for in the temporality question, the space of a sound is not its spectrum; that's perhaps a space of a sound. We can distinguish an enveloping or ambiant figure from an object (gegen-stand) figure, this would be the architypical too ends of spatiality. An enveloping figure could be a deposition in the space, probably a conditioning in Minard's sense, or it could be transparent in the sense of a Cage-like unimpeded pentration; likewise, an object sound could either articulate the existing space (focus on particular properties, work them out, exaggerate them), or it could stress its independence.


{function: response, keywords: [temporality, spatiality, sound]}

I think that mimicking / echoing could be one strategy, but among others for the articulation in the space. One is faced here with a mostly unidirectional flow of information, such as the steps of people which one could use to synchronise the machine, but it might be hard to assume that people "lock" into emitted rhythms. Surely, one can create irritations and slight shifts, "bends", which make me think of the situation where you walk along with other people, and you find yourself either matching the pace of your steps with the others, or clearly distinguish them (a form of synchronisation or desynchronisation).

 

A more autonomous behaviour in terms of temporalities could evade this "control" situation. How could the behaviour be both informed by the sounds picked up from the space and at the same time remain simultaneous with their ongoing?

 

{function: response, keywords: [patterns, interaction, coupling, mimicking, synchronization]}

I would say exchanging a channel of sound signal is neither viable nor desirable (as a form applied to all communications in our systems). Indeed, is not this signal channel a "simple first order" communication? I don't find it desirable, because it does not require a priori discussion of how the communication could be structured, in the bidirectional sense I was mentioning in the response to Daniele. It feels akin to saying "here is some chunk of 'sound' or data stream, use it in whatever way"… It is probably also not viable because a) we would have to introduce a lot of extra bandwidth between low computation power units, also new sound interfaces, and b), more importantly, it would rely on the idea that each layer can produce such an additional 'sound' layer distinct from the sound it outputs, which may or may not be the case. So I don't want to rule it out (although practical aspects speak against it), but it would depend on the bidirectional connections we are forming between each two layers.

 

{function: response, keywords: [rules, communication, protocol]}

It could even be an operation nested within an operation, before we even think about the linearised time coupled to audible sound.

 

{function: response, keywords: [operation, time, sound]}

I would not settle on an either-or in the second question. The two poles would be complete "containment" versus complete "artificiality", with the word "deviation" always having a positive gravitation towards the equilibrium (the deviation is defined by the equilibrium, so it is never truly in-itself).

 

{function: response, keywords: [deviation, equilibrium]}

materials 16-dec-2019

Same process with a public domain sound recording of a person walking a staircase.


{kind: caption}

Resonance experiment (sound only) by JYK.


{kind: caption}

Auto-correlation at fixed spacing, DC blocked.


{kind: caption}

Segmentation (right) based on local minima in input (left) auto-correlation.


{kind: caption}

Segmentation and subsequent minimum phase transformation. I had some trouble in the last segment, which I guess is some bug that can be found.

 

{kind: caption}

I will now try to develop some materials relating to these questions and responses.


{function: contextual}

Corresponding sonogram. This simple method works more or less; the first cut seems a bit off. I also noticed strong difference between left and right channel (I'm using minimum of both channels now); perhaps there is something to do with very low frequencies affecting the analysis. Probably we should add a perceptual correction filter first.

{kind: caption, keywords: [analysis, signal, spectrogram, group: seg-sono]}

Corresponding sonogram.

{kind: caption, group: seg-sono2}

Thinking, for example, of a "dictionary" of foot steps, and Ji's idea that there could be a connection between the first and last step within a gesture, this could also mean to respond with a memory of steps.


{kind: note}

I would like to work with simple statistical properties, for example the segmentation in the simple examples is given by the most pronounced local minimum in the measure, within a predetermined window (elasticity). I would like to obtain these windows from histogram analyses.


 


{kind: note}

How to open a Mellite workspace

Make sure you have the latest version of Mellite - you can download it from here - instructions are here. Read the instructions also for information on SuperCollider installation.

If have extracted the Mellite zip download, you should be able to launch Mellite from the bin directory. It takes a moment to come up. Once the menu appears, you can then open the workspace that you have downloaded separately (extract the .mllt.zip archive to obtain the .mllt directory/database). Select File > Open and go into the parent directory of the .mllt workspace, here select the .mllt directory (do not double click), then select the Open button.

See this short video tutorial for a demo on how to download Mellite and open existing workspaces.

{keywords: Mellite}

---
meta: true
author: HHR
artwork: ThroughSegments
project: AlgorithmicSegments

keywords: [questions, responses, proposal, brainstorming]
---