Here we discuss about the conception phase of the project.
The concept of the composition may be a core idea that stimulates creative thought. Could be a poem, an image, a sound, a philosophical question or anything else you can imagine. This concept will guide you to select and process the initial sounds (recordings), and develop a dramaturgy for the composition. It doesn't have to relate to machine learning, since the ML algorithms will not be aware of so high level ideas. The ML algorithms will assist the creative process by learning some features of your composed fragments and generate new material. It will be up to you then to further edit and contextualize these generated fragments.
There will be also some limitations concerning the electroacoustic composition techniques. This is done like that in order to facilitate the operation of the ML algorithms. We are going to work with the Reaper DAW, which is available for all operating systems and free to use for the purposes of this project. You will be able to work with as many tracks as you want but we have to decide together (in a later time) a fixed number of FXs, which will be allowed for use. It is also good to know now that the composition will be written for two output channels (stereo).
This discussion until Friday is an opportunity for you to discuss any concepts you may think of. You can get early opinions about your concepts and see if you would like to dig further or develop new ones. I will follow the discussion and intervene if a concept may fall out of the scope of this project. Please feel welcomed to share any thoughts and questions!
I was the last at the introduction part, by I can start the conception phase here.
It is not easy to give an idea that will find an approval of so many people in a group. Especially if we're going to work several months developing, correcting, changing and enriching our own products.
I will not write about any process here, as it's not the part of a conversation yet.
Let's focus on the ideas that we would like to present in a new composition collaboration. My suggestion is:
it could be great to show in the project the aspect of collaboration as a creative element - through a statement, maybe a philosophical phrase, a poem - could be recorded in sounds or in speech; my suggestion is to use our own voices, or even a native language of each of us, to state in some of prerecorded materials. By that we could be present not only as a composers, but also a vivid part of composition itself, that will be transformed through the agents and other procedures. I can think of some poem examples, but first need to know your opinion and other options.
It's just a beginning of a brainstorm, I'm starting so we don't get shy and reach to some conclusion till Friday 🙂
Hello Martyna and thanks for your idea, it sounds very interesting!
I agree that it would be nice to reflect the collaborative element, but also how this collaboration, featuring both humans and artificial agents, might cause the emergence sort of 'meta-composer', or 'meta-consciousness'. I had an idea today, so I'll just present it below before I forget it:
I was thinking of Gilles Deleuze and Felix Guattari's concept of "Double Articulation". I won't be able to do it justice here (nor am I an expert on this), but a rough interpretation that might work for us is the idea that everything is shaped by the self-organising process of 'double articulation'; the first articulation is 'content' and the second 'expression'. And this process creates 'strata'. About strata, they write (in A Thousand Plateaus),
"[Strata] consist of giving form to matters, of imprisoning intensities or locking singularities into systems of resonance and redundancy, of producing upon the body of the earth molecules large and small and organizing them into molar aggregates. Strata are acts of capture, they are like "black holes" or occlusions striving to seize whatever comes within their reach."
And about 'double articulation',
"The first articulation chooses or deducts, from unstable particle-flows, metastable molecular or quasi-molecular units (substances) upon which it imposes a statistical order of connections and successions (forms). The second articulation establishes functional, compact, stable structures (forms), and constructs the molar compounds in which these structures are simultaneously actualized (substances). In a geological stratum, for example, the first articulation is the process of "sedimentation," which deposits units of cyclic sediment according to a statistical order: flysch, with its succession of sandstone and schist. The second articulation is the "folding" that sets up a stable functional structure and effects the passage from sediment to sedimentary rock."
So this example reflects how geological processes bind large collections of objects or stuff into general forms (first articulation) and then render or fold these into rocks and mountain ranges which are like 'frozen' dynamic movements (second articulation). But the general idea is that collections of things together become new emergent expressions – not the expressions of a single agent, human or otherwise, but the expression of the process itself.
So I was thinking that the process of composing this work might be somewhat like a long succession of double articulations: we contribute with sounds, we refine the sounds, we aggregate the sounds, we shape the aggregates, we select composed sequences, the AI composes with our sequences, we articulate those further etc, and so on. It's like several 'acts of capture' or strata, created over the course of the process, always based on a collective agency. Of course, there is no essential distinction between the 'human' and the 'technological' contributions in this sequence. Training a machine learning algorithm and having it generate output is perhaps like a kind of double articulation in itself.
I'm not sure how exactly this could guide us in terms of what sounds to create, but one possible way of looking at it (which is very open) could be to just say that we all contribute with material that is distinct but also has further potential for shaping. For me, what comes to mind is stuff like textures, micro-sounds, ambiguous or disorderly sounds etc. But that might not be what another composer wants to contribute with, which is fine – each individual composer's contribution could (and probably will anyway) reflect their aesthetic interests, as long the materials are not too refined already. As such, I imagine it could include voices and languages, as Martyna suggested, or anything else. Then we could compose by looking for connections and differences between these materials and somehow let that dynamic give way to new sounds. The 'meta-composer' (all of us and the technology together) eventually synthesises all that stuff which originally may be quite individual into something new which we don't know about yet.
So that was an idea. I hope it makes some sense. Needless to say, it's open to be stormed upon or corrected by your brains 🙂 There are probably better ways to put this idea in practice that I haven;'t thought of.
Sorry about the really long post!
All the best!
Following Martyna's and Erik's posts, that I like, I may try to synthesize a bit the two of them. I think having a theme at the highest level could be helpful. It could be a title, a poem, an idea, an article, a painting, etc. this could be related to concepts such as diversity, sustainability, privacy, singularity, etc. or something more specific objects/ideas such as rock, balloon, creak, friction, rumble, etc. but this might already be too restrictive... Perhaps the MLA could be involved in this process as well.
I think the double articulation idea of content and expression and the process of strata could allow us to have a collective mindset and at the same time, some flexibility.
I would also like to see how the MLA analyses and synthesizes our ideas/sounds and sort of challenge it. Then see if we need to articulate our collaborative process further.
I have added a few more steps/questions to the ideas already mentioned by Martyna and Erik:
2- duration? (8-12 minutes?)
3- contribute with sounds (how many or how long?)
4- refine the sounds (MLA will help with this?)
5- aggregate the sounds (perhaps the MLA should be involved in this step as well?)
6- shape the aggregates
7- select composed sequences
8- the AI composes with our sequences
9- articulate those further
10- refine the overall shape of the piece (should we try to draft an overall shape of the piece at an earlier stage or we deal with it later as we evaluate the potential of each aggregate or sequence?)
Thank you for your thoughts and questions! It seems that you are trying to interconnect the multi-level ideas that constitute this project, which I think is a great approach. We tried to design a very general collaborative framework so that it can work with composers of diverse background and with any sound material. The main idea behind this design is that the composers (as well as the ML algorithms) function as nodes in an interconnected network. Each node may have its own aesthetic preferences and biases but through an intense information exchange, a collective compositional consciousness may emerge that transcends the creative capacity of each individual node (this is one of the project's research questions). So the more information is exchanged between the nodes (you!), the more the network operates as a unity and maybe this goal can be achieved.
The "Double Articulation" description and the geological processes are very relevant to the process we are going through. I think that the traces of these processes will be inevitably imprinted in the final composition, as all geological processes leave their mark on the terra firma. So, I believe any theme-concept-idea that will guide you in finding sounds and develop a dramaturgy for the composition will work. You can propose also ideas that are recurrent in your own compositional practice but try to articulate them in a way that is understandable and attractive to your fellow composers.
Today Friday is the last day of the brainstorming. At midnight you have to submit in this webpage your concept. Each one of you has to submit one concept. In the weekend I will prepare an overview page with all the concepts where you will be able to arrange them with the order of your preference. On Monday I will announce the most preferred concept and the preparation phase will begin. Needless to say, you will be able to further discuss, modify and refine the selected concept in the subsequent phases.
About the more concrete questions by Panayiotis, I have already some answers but I would like to reveal them as we go through. Some of them may of course change according to the selected concept and your invaluable feedback.
I am terribly sorry to be late! I was on my way to the US which, due to delays on flights took much longer than expected…
I read the posts and I also like the idea of somehow including our voices in the pool of materials, this could also form the basis of our initial material! I also encourage that the abstract concept of 'collaboration' as such could lead the way in the artistic work. As I understand the output of the ML algo will be strongly defined by our input material, in a form that it will try to 'predict' some new fragments based on our input fragments' various features. I could frame this already as a form of collaboration, ie. it is trying to give me "more of the same" in a way, but of course, it won't be able to do that perfectly, and the error of the prediction itself will make things artistically interesting and lead to the emergence of new stuff. Erik, as far as I understand, this is somewhat in line with the framework you proposed as well, right? I find this somehow relevant to our collaboration too, in a way that we try to collaborate with others based on our artistic experience and taste, but inevitably there will be 'misinterpretations' of each other's intentions or even styles, which will add "anomalies" into the mix, hopefully emerging into something unexpected (and cool).
Just brainstorming further: I find the idea of "misinterpretation representing/giving way to new value" interesting, or how imperfection/error being the main source of 'fortune' (i.e. interesting results), as being somewhat poetic as well (and also very relevant in the field of ML art). Maybe we could define some workflows in which we use the ML algo and perhaps each other's material as well - individually - to generate a pool of gestures/textures which we can later stitch together (defining the form/dramaturgy/time structure) as a team? Through the lens of Erik's idea, there would be a stage where we create content, and a "second articulation" where we try to put the given gestures/textures (maybe without knowing who made them originally) into some shape, add something we might think is missing, or transform into something which we think is interesting. This might or might not be in line with what the author of the original gesture/texture had in mind. We could even spice this up by for example giving the same set of material initially to all of us (like samples with our voices), and then when we "lace" our individual results together, perhaps emphasising the variety of results in the final structure in some way. Maybe I am going a bit far with this - talking about the work process, but just take it as a mere brainstorming. The core concept of this would focus on the role of mis- or reinterpretation in creative collaboration
Hello all! I've been following this thread very closely and will have the PDF with my thoughts before midnight. Although it seems that we are reaching sort of a consensus in terms of the conceptual framework. @kosmas , should we attach a pdf to a new message here in this thread?
Hi everyone, in this webpage (link) you can order the concepts according to your preference. You are not allowed to choose your own concept. Please submit your preferences until 13.08.2019 at 23:59 in your local time-zone.
Hi to all! All your preferences have been recorder and plotted in the attached histogram. For example the concept 1 received 0 first preferences, 2 second preferences, 1 third preference and 1 fourth preference. It is clear that the concepts can be arranged according to the group preference like this: concept 4, 3, 5, 1, 2.
Hence I would suggest to start the preparation phase based on the concept 4. There is also the possibility to merge for example the concepts 4 and 3, if they could be compatible. Or we can extend the concept 4 with ideas from the other concepts if they could fit in. We can discuss that shortly until tomorrow night but we have to start the preparation phase on Thursday.
Thanks to everyone for their great ideas and to Kosmas for counting up the votes.
Though it appears that my concept won the poll, I want to say that I don't think that should exclude the other contributions and that there is scope for blending the ideas. As Kosmas alluded, I certainly think 3 and 4 are compatible. But there were great elements in all the contributions, and it seems that we've already been influenced by each other's ideas which is great. Anyway 'double articulation' is, in practice, quite open for interpretation and can easily be viewed from the lens of the other ideas. There were some recurring themes that I think we could use. E.g. collaborative authorship - misinterpretation could be seen as a form of articulation or expression in how it brings forth new meaning. Emergence – collective agency as a crystallisation or morphological poesis. Questions concerning the human and the non-human / post-human – if the 'content' is human voices, then the 'expression' might be a kind of posthuman meta-voice. Just some thoughts.
What do we do next?
Hi Erik, thank you very much for your message. We have entered the preparation phase now, so I would suggest to continue the discussion in the under the preparation tab.