Reverberation

I tried out Daniele's modulating delay lines patch, converting it to Scala. First, I made a mistake in the delay lengths calculation and multi-channel-expansion; but this also sounded very interesting (the number of lines is squared compared to the original :)

{function: contextual, date: 191218, keywords: [delay, delay lines, Scala, SuperCollider, translation, FDN]}

Corresponding Mellite workspace

{kind: repository}

Circularity

Ji:

I imagine a circular process, where a certain process can be repeated without having a fixed set of duration, but depending on the external circumstances ↗

{kind: quote, persons: JYK}

Influence

Ji:

forcefully influence the room resonance

{kind: quote, persons: JYK}

Transformation

Ji:

But then I'd probably add some sense of segmentations here (how can we combine those two contradictory concepts?) thinking of Lucier's ways of transformation: meaning introducing the time domain. ↗

{kind: quote, persons: JYK}

Topology

Ji:

instead of thinking too much of the concept of 'division,' we can think of an open set in a topological space.

{kind: quote, persons: JYK}

…Topology/Bridges

Ji:

I found this (Petri nets) very interesting/inspiring. I wrote the text about bridges below DP' text and then took a look at this, then somehow what I've imagined was there. Can we (Would we) integrate this somehow? Not necessarily exact to it, but we can take the idea, and modify the way we'd like? ↗

{kind: quote, persons: JYK}

{hhr, 200124}

I would also like to work without any fixed set of durations, while at the same time working with durations "at all". That is, I want to derive segmentations in time from the analysis of the incoming sound signal. I then imagine that a set of processes operate on these segments, they succeed at later points in time (depending on the time it takes to process, thus on the speed of computation), and then they inject their products into a set from which the loudspeakers are fed. So all processes are repeated with intrinsic durations, and the coming into the audible of their products will be coordinated somehow by another process, that makes sure that "slow" processes are not disadvantaged against "fast" processes, and certainly based on a compositional decision on how the proportions could work out.

{function: comment, keywords: [time, segmentation, duration, loudspeaker, analysis]}

{hhr, 200109}

@Ji - do you have any pointer here, where Lucier explains transformation?
{hhr, 200110}
It was meant to refer to the way I'm sitting in a room is infolding in terms of the form given by the space itself.

{function: comment, keywords: [space]}

{hhr, 200124}

For now I will suspend my looking into Petri Nets. I find the the concept of transitions and places interesting, but I have not found, so far, a way to enter PN as something I could employ in my composition. Most papers are interested in formal analysis of systems (e.g. reasoning about concurrent systems), or building optimisation systems. I still think it could be interesting to work with them, but I don't see a realistic window for this project to dive deep enough to make anything useful out of them. The original paper (dissertation) by Petri might be an interesting examination at some point, it has many diagrams and drawings for the development of the thoughts.

{function: comment, keywords: [petri, net, transitions, places]}

Scales

DP:

Spatial (/ temporal) relations might repeat, similar, along different scales, a fractal of relations, extending through dimensions and scales.

{kind: quote}

{date: 200108}

In a further step, translating from time domain to Fourier domain, such that clicks become pitches.

{function: comment, keywords: [fourier]}

{date: 200108}

I used in a piece back in 2016, "Accel", a simple process of accumulating a microphone signal from an outside space, downsampling it by a factor of 64 (as far as I remember). I had also used that before when I was working the Rondo residency long time ago. It's very interesting, because new rhythms emerge from the traffic and heavy machineries, while human voices almost disappear in the high frequencies.

Here is an experiment with one of the staircase field recordings, accelerating it by factors of 16, 32, 64 (this time using band-limited interpolation).

{function: comment, keywords: [microphone, accumulation, downsampling, rhythm, traffic, machineries, voice]}

16x

32x

64x

Corresponding Mellite workspace

{kind: repository}

compare

{hhr, 200109}
Foreground/Background

Working on the foreground/background question, I tried without success to apply the dereverberation algorithm to the test recordings; probably it does not work well, because it needs multiple channels (and my implementation is currently broken except in monophonic mode), also the rendering speed is way too slow.

So the next idea was to threshold the spectrum somehow. I ended up with two sliding percentile filters over the short time spectra's magnitudes, creating thus an adaptive filter. Here is an example from the field recording in the staircase, original and 32x speed up.

{function: contextual, date: 200109, keywords: [foreground, background, dereverberation, filter]}

Corresponding Mellite workspace

{kind: repository}

{hhr, 200124}

Or perhaps not eight individuals, but two—the two rails, which could carry out counter movements, and the four channels of each speaker being somehow in the same space / alignment / orientation. That would still offer the possibility of heterogeneous combination.

{kind: note, author: HHR, date: 200123, keywords: [discreteness, segmentation, speakers, rails, mixing, analysis, clustering, foreground, background]}

Discreteness/Segmentation

I'm thinking that I like to hear the 32x sped up sound along with the slowed down extracted resonances I put on 11-Jan-2020 on my 'pin' page. And it gives me the idea that I should treat the eight speakers individually and not mix these different treatments on the same channel. I wonder if I can come up with eight different treatments / type of analysis and segmentation, thus?

Although I'm interested in the vagueness between foreground/background, this is perhaps the opportunity to put them in opposition (one rail against the other). Thus: Can I find a counter-movement or complementarity for the two existing approaches, and for two more approaches to be determined.

Another strategy not yet explored is sorting (ascending, descending steps) of segments. I have used that before in various ways, and "removing the syntax" can yield very interesting things. / by extension: clustering (although the 1-dimensionality has a nicer relationship to the stairs)

{kind: note, author: HHR, date: 200123, keywords: [discreteness, segmentation, speakers, rails, mixing, analysis, clustering, foreground, background]}

{hhr, 200124}

Because I started reading this again, I thought I could share with you the first chapter of Sara Ahmed's book from which I took the quote about the simultaneous arrival. It's an examination of phenomenology and how one can write about it. I don't propose that we all read it and/or talk about it, but just to give you something I'm looking at, perhaps you are interested in some moment.

{kind: note, author: HHR, date: 200123, keywords: [discreteness, segmentation, speakers, rails, mixing, analysis, clustering, foreground, background]}