Hi all

I'm not sure that this post constitutes a "technical issue", but I'm posting it here anyway.

I took up PD sincerely this year, after over five years of flirting with it. I've loved electronic music all my life and always wanted to create it myself, but could never work with any of the existing suites (Ableton, etc). PD seemed like the right choice for me, partly because it is elemental and versatile, and partly because it is materially and ideologically free. In just a few months, I've grown comfortable and even competent at patching, and would count it amongst my greatest hobbies. I love the sense of control that PD allows, and the way in which it merges programming with design. I even sort of love some of PD's quirks, ("trigger: can only convert 's' to 'b' or 'a' " etc.). I've made several patches which I am proud of and have learned a great deal in the process. But there is one problem that keeps on nagging on my mind:

My musical output has been almost zero!

I have created all sorts of weird and wonderful contraptions for creating music, but very little actual music. The trend is pronounced enough that my wife has affectionately coined the term pre-music for my craft. But as the trend continues, I worry that all I will ever produce is more pre-music, more and more elaborate but never musical.

I see two problems with my current condition, one more technical and one more artistic. I would seek advice or reflection of any sort on both.

1. I have gotten pretty good at the dataflow and control elements of PD, but am less skilled at sound design. I know how to make basic synthesizers and samplers, but I struggle to create sound that actually sounds good. I'm sure that this is the kind of thing that comes with practice, and that I should just start constructing more advanced synthesizers, but I still often don't know where to start.
2. Even with a functioning, sonically appealing program, I still struggle to make music. It's that feeling you get when someone who plays an instrument beautifully places one in your lap, and you don't know what to do. Your creativity ceases up and you just feel clumsy. I know how to play some instruments, but I still don't know how to make music on computers. How do you access your creativity on a machine?

I am guessing that many others have faced similar challenges to some degree of other, and would value your advice or discussion.

Actually my advice is the following : don't be afraid to sound bad. The best musicians don't try to sound good. In my opinion, good music is not about good sounds. Beauty is not something pleasant.

The best musicians don't try to sound good. In my opinion, good music is not about good sounds. Beauty is not something pleasant.

This is absolutely spot on.

the BIG problem with PD, in my opinion, is that there is no native way to set up state saving.

If you could patch a bunch of stuff, and then press a single button to save the state of all your sliders and number boxes and whatnot, and then be able to easily manage all those things as 'patterns' or 'songs', it would be downright ecstatic.

Miller has said for years that he wants to implement this, but he never has. Ironically, he keeps spending time on stuff like the data structures, which are clearly a dead end and barely anyone uses cos they're clunky at best and ridiculous at worst.

If PD had a simple, fairly user friendly way to save and recall different settings of patches, it would suddenly open the door to far more complete and coherent productions.

meanwhile we all just hack away using stuff like SSSAD which is OK, but not ideal.

And here is an example about beauty vs. sounding pleasant from the music that I deal with the most: almost all pieces by Pierre Boulez which have been written from around the 1980's are easier to listen to, and have a very sensuous and attractive kind of sound (such as Dérive 1 or Sur Incises) but are compositionally poorer than any of his early masterpieces (Structures 1a, Le marteau sans maître, Piano Sonata No.2), which are tougher for the listener, rougher in sound, etc. I also would take any piece of Xenakis or Stockhausen over those late works of Boulez any day.

@mod I don't understand you, what is wrong with using a message containing several sends, such as [; myReverb 1.0; delayFeedback 0.8;, delayTime 250; structureB bang; soundPlayerA 0( which can be banged by an [loadbang]? Doesn't that initiate or change the state of almost everything in a patch?

Take care,
Gilberto

@gsagostinho I totally agree about Boulez. Beethoven did the exact opposite : he evolved to the harsher sounding. I think the best example is Boulez's choice of composing his livre pour cordes for string orchestra and not for string quartet.

@mod Maybe you should just make sure you have a simple access to every parameter. There are several ways to do so ; as for me, I use the "receive symbol" (not sure it's "receive symbol" in english) of almost all the GUI I use, so as stated by Gilberto a simple message box can set many parameters at once, and it appears visually on your patch. Maybe your problem is that you'd like to be able to store a set of values you've found playing with your patch, but it's not really a problem. You can store in real time all your values, then save any given combination (in an array or a list or a text file) so you can access it any time.

@gsagostinho

information from outside tends to affect me very little (actually, it can in fact take my concentration away from these musical structures).

Interesting glimpse. But have you ever considered building some bridges? Bring to your mind the sound of rain, for example. Isn't nature filling your mind with random sounds? Listen to yourself: "there is very little that listening to the nature or life around can bring". There is something curious here...how can a musician feel that listening can bring little? Isn't listening important?

So this is my proposal : next time it rains, you listen to the rain as aware as you can...every detail, small sounds, close sounds, far sounds, the roof, the wind maybe, somebody trying to get a taxi, thunder. Then read your own phrase: "there is very little that listening to the nature or life around can bring".

@Load074

I've always been nuts for the classical tape music folks

You mean people like Daphne Oram? I love that bitch!

@mod

Miller has said for years that he wants to implement this, but he never has.

I think that somewhere in the documentation it reads that the idea was discarded or sthg like that. Here I found sthg, although it's pre 0.35:

In Pd there's no "preset" object (I now think it's basically a bad idea) and you have to use explicit sends and receives to restore values to number boxes. Then just make a "message" box to re-send the values you want.

@gsagostinho said:

@mod I don't understand you, what is wrong with using a message containing several sends, such as [; myReverb 1.0; delayFeedback 0.8;, delayTime 250; structureB bang; soundPlayerA 0( which can be banged by an [loadbang]? Doesn't that initiate or change the state of almost everything in a patch?

Take care,
Gilberto

I think setting up the savestates is a bit awkward, myself. I think it was Obi or Brinkmann that made a nice one-button-save. I dunno if it's just me, but saving data in arrays (Specifically for the cool wave shaping distortion array) doesn't seem to work. It's mostly OK, because I set up [loadbang] automatically at this point, but it would be a very handy piece of functionality.

@LiamG , which standard DAW do you use? Can I suggest using the MIDI functionality with MIDIYoke to send data from PD to another DAW? I find using PD to compose while I use something else to handle the sounds to be my preferred method of computer music.

@LandonPD Hi Landon, it is not as if I have never heard of composers using natural sounds or influences to compose music, and when I was young I was certainly much closer to that way of thinking - so it is rather myself destrying the old bridges rather than any need to build new ones.
A composer can certainly lack interest in every sound around him, as much as a writter has no interest in every single word written around him. My own work is about sound relations an structures rather than the sounds themselves, and that is not a new approach at all. There are composers and schools of music in which sound itself is the ultimate goal (Pierre Schaefer and his acousmatic music, or the spectral composers), but others approach music in a different way. One of the reasons I love so much randomness is that I can't bother to chose myself between certain details that I don't judge to be important.
That is not to say that I have never felt attraction to beautiful sounds around me: for instance, I have a fantastic metallic bowl which has a prettier sound than most bells have. But that said, I have very little interest in recording and using it in my main work. I distinguish these two things greatly: one can admire a landscape and yet not paint one nor be influenced by it at all when painting.
So as much as it is possible to compose music by listening to those sounds around us, it is also possible to do so by NOT listening to those sounds either, and the proof of that is every single work which has been constructed only by musical relationships.

@What I agree with you that it could be easier, but I don't find the current way of dealing with it that awkward myself. But anyway, I don't know this state saving object you mentioned, do you know where I could find it?

Take care you all,
Gilberto

Hi, tossing in some cents:
in my experience my issue with Pd is the opposite from LiamG's issues: the sound design is straightforward and simple to implement (again, depending on what's being implemented) but the process of algorithmic composition is not as well suited to data flow programming. My main interest in computer music is algorithmic composition and I believe pd was designed as a tool for live performance which makes it unwieldy to use for algorithmic composition but great for what it was designed for. Control of events over time is severely lacking in my opinion

Let me give examples:

1. There is no way, at a given point in the program flow of a patch, to call a subpatch with certain input arguments and get the output in a certain format. Instead, copies of the subpatch must be made, increasing memory usage compared to the classical idea of a function.
2. When composing, I like to think in terms of "do this thing at this point in time". Say I have a melody and at a certain point in that melody I want to choose between 2 different notes. One way of doing this would be to have a qlist that plays the melody and have it call a receive object that takes parameters to choose between 2 notes. But what if I want to call that function to play a different instrument? all of a sudden I need an extra argument to that control structure, and need to re-program the subpatch in order to do it.
3. Say you want to quickly edit a score of midi notes or midi automation, which maybe don't control midi directly but rather control algorithmic parameters that change over time. This is very difficult to do. You could set up a DAW to send the midi to pd, but MIDI is not designed (in it's structure or in it's output format) for sending data to programming languages, but directly to instruments. There aren't many open source OSC editors either (although there are a couple of promising ones).

recently I have been exploring other programs for algorithmic composition purposes, such as Common Music, based on Scheme. What makes Scheme enticing for algorithmic composition is that it supports continuations natively. What this means is that you can control the program flow manually, and the score for an algorithmic composition could be set up to be the program itself. (Common music does not support this but it could be set up to).
So, in the example above, when composing with Scheme set up in this fashion, all that would be required to choose between 2 notes and play it would be 1 line of code, like (play myinstrument (choose 49 64 20)) and to play it on a different instrument all that needs to change is the first argument. Notice that "choose" does not need to be rewritten.

Personally, I believe what is needed for the future of algorithmic composition is a dedicated DAW-like environment, (I have been thinking of undertaking such an endeavor). It would have tracks, and each track could have several layers that could output to either OSC, MIDI, or another layer in the project. Layers could be midi note editors (that could be configured to output OSC), automation editors with configurable precision (like if they output ints or floats, and how it is scaled), pieces of code that generate notes, pieces of code that take input data and output other data, containers for more layers, etc. The programming languages used would have to bridge over a common interface somehow, and they would need to support continuations or coroutines (Lua does also). The editors could have functions to, say, bind a specific midi note to a specific function in a specific layer. This would combine the ease of using DAWs to compose temporally with algorithmic composition constructs in my view.

But all that is just based on the way I think about composition... everyone's different.

@gsagostinho Turns out both Obi AND Brinkmann have save / load systems.

Obiwannabe's Six Simple Synths and Martin Brinkmann's fmdrone 1. Brinkmann's patches are always a huge mess to look at, though. I always wondered how he manages!

@seb-harmonik.ar hi Seb, did you also consider Super Collider? As far as I know, it has a much more linear way of dealing with information, and you can use functions there as well. I tried to learn it a couple of times but never had the energy to really go through with it...

@What Thanks a lot, I will have a look on them soon! (EDIT: I had a quick peak over Brinkmann's patch, and that is the biggest monstrosity I have ever seen in Pd! I am so neurotic with my own patches that I try to keep all objects having the same distance between each other and all cables as straight as possible, so you can imagine what that patch of his can do to someone as compulsive as I am...)

@gsagostinho that's a good idea, I've tried to learn SC before but I'd forgotten about Routine (a control structure that supports coroutines). It's probably the best option for what I describe in terms of algorithmic programming but SC still doesn't have a good editing interface as far as changing the parameters with time (e.g. midi automation)

Perhaps it would be easier to write a plugin for SC than to build something from scratch though

@What and what do you think about this super simple solution to state saving? http://forum.pdpatchrepo.info/topic/8782/state-saving

@gsagostinho said:

I love so much randomness is that I can't bother to chose myself between certain details that I don't judge to be important.

Then you are going to like this article about a pattern discovered in the 50s. http://www.quantamagazine.org/20130205-in-mysterious-pattern-math-and-nature-converge/

@LandonPD hi Landon, thanks a lot for the article, it is indeed very interesting! I am currently writing my master whose theme is randomness and algorithmic music, so these type of stories are always very inspiring to me

Awesome. I wish I could read it some day.

@LandonPD Thanks Landon, I would be more than glad to send it to you once I have it finished. I actually still have more than one year left until my deadline, but since I want to study abroad for a couple of months close to that date I do plan to finish it much earlier.

@LiamG Hi Liam, one more suggestion: if you are looking for a nice free synth to produce sounds, have a look on ZynAddSubFX (http://zynaddsubfx.sourceforge.net/download.html). It used to be only native for Linux but they also released a VST version of it, and it has some very cool sounds.