hello,

i am about to implement about 20 nearly identical data processing things i've built, and i'm wondering if i should insert them as subwindows or abstractions. they are meant to run in the background so my only consideration is speed. is there a significant advantage one way or the other? anecdotal, non-scientific answers welcome.

leo

i am trying to create filenames in pd with 4 or 5 variables and one constant. they can't have any spaces in them or soundfiler won't be able to read them. this presents some problems because [makefilename] only accepts one variable and dollar-sign variables have to have spaces in between them.

i'm stumped. has anybody done this sort of thing before? any suggestions? are there any other objects i should look at?

thanks, leo

does anybody know where i should go to work on the documentation for pd? what about documentation for externals to pd?

leo

hello,

i am making progress on my drum machine. i now have a working version with no bugs. there are many next steps, but i'm concentrating on the issue of saving patterns because it is the most mystifying to me. right now, i can load beats using qlist.

the sounds are controlled by a grid of toggles, each row representing a drum sound and each column representing a beat (there are sixteen beats). when a toggle is on (1), the drum sound plays. when it's off (0), nothing.

from my point of view (one of ignorance when it comes to the qlist and textfile objects), the most obvious method of saving patterns would be to save each change as it is made. since that would only alter a few characters in a textfile, i don't see it as using too much memory.

note: as i wrote that last paragraph, the "obvious method" changed from something very stupid to what you see. i think i can figure out what i can described without help, but i would still appreciate any ideas you have.

leo

i am trying to figure out how to read audio from a delay line in reverse. clearly, it can't be done as simply as that line suggests, because the delay isn't a static array, but is merely playing the audio input through at a later time. it would need to play into an array instantly when the reverse feature is switched on and then read the array backwards without any significant latency. i can sort of muddle through a bad solution based on the samplers included in the tutorials, but those are absurdly resource intensive and cause a lot of audio dropouts for me.

if anybody has any ideas on how to do this more simply, i'd appreciate the help. it may be that i should just buy a delay pedal, but they are so expensive.

hello!

i am trying to build an auto-switching tool into a phase modulation patch i'm working on. i want it to use the amplitude of the input sound as the modulation index when the manual index control is set to zero.

right now, i'm trying to work out a method using spigot, which lets streams of numbers through the left when the right input is nonzero. how can i turn the manual index control numbers into zero when they are zero or negative and let them be positive when they are?

is there an object designed for this, or does anybody see how it could be built?

boolean logic would make it easy; is that available in pd?

thanks in advance,
leo

i want to feed adc~ into a phase modulator. right now, i'm using fiddle to take it apart and reconstruct it as a set of sine waves, and it sounds very watery. which is great, but i want it to sound like normal sound.

i'm using the help file about phase modulator, which takes numbers as its input, and i haven't figured out how to feed the audio in yet. any suggestions?

leo

another question about the same patch:

i know that multipe instances of the same send~ cause an error, but i have eight of them and they all seem to be working. does anyone know why they cause an error and any problems it might cause that i can't see?

because right now, they seem to be working perfectly.

leo

hello!

i am very new to pd (only been with it for a couple weeks) and am trying to figure out how to take an audio signal being played through the soundcard by another application or by the OS and split it into three (or any number) of frequency ranges.

i want to use this to take an audio signal, split it into three ranges, and convert each into a sine wave that would play at the average frequency of its corresponding range, taking a new average very frequently. my friend did this in max/msp and i'm trying to replicate as an exercise and because it sounds amazing.

i'm only looking for help on the first part because i want to try to do the rest on my own. we'll see how that works out.

thanks for any help, leo

hello,

i need help playing back some samples. i want to be able to start playing a sample while it is already playing, without interrupting the first playback. any ideas?

leo

hello,

i want to use a counter that runs from 1 to 16 to send bangs to my receives, which are named monitor-1, monitor-2, etc. is there a way i can add the the number to the "monitor-" and then send?

i could just name the receives 1, 2, etc., but i am going to have different sets of receives all receiving bangs in the same manner, and so i need to differentiate between them.

well, it might be possible to name them all the same thing and only use numbers in the name, but i'd rather not.

having a version of pack for messages would be ideal, as long as i didn't need a space in between the messages i was packing.

thanks,
leo

so i misunderstood your problem slightly (i just skimmed and thought you were saying that everything was displayed at the wrong size and so objects and words were all jumbled up). a little trickier. those other options would work here, but a better solution (at least until you move to a different version of pd) is to open the patch in a text editor and change the dimensions yourself. then the help patch will always open at the right size. you will need to ctrl-click on the pd package, and display the contents. go to resources, and then doc. here are all of the help patches. open the one you want, and look at the first line. here is A01.sinewave.pd's first line:

#N canvas 6 2 588 513 10;

the '588 513' is the dimensions, in pixels. width and height. you will have to change the permissions for the file and the directory in order to save the changes. kind of stupid and harder than it should be.

here's my workaround. hit ctrl-e or cmd-e to get into edit mode (i don't know what you press in linux or irix), and then either select all and copy it a new window or just select and drag everything around to simulate scrolling. you will need to do the latter if the patch involves a sample, because the sample won't get copied into the new patch window.

what i meant was that i want to contribute to the documentation. there are too many undocumented and poorly documented objects in pd.

okay,

i ran a test where i made a patch containing 10 instances of the following code, once as abstractions and once as subwindows. this code runs as two [inlet~]s, hooked up either to a [bonk~] or a [fiddle~], each of which is then connected to an [outlet] through the left-most outlet. besides the load meter, i had no other patches open. i left dsp on for 60 seconds for each of the two options. based on this load meter, i observed no difference in processor usage. on my ibook g4 running os x 10.4.5, the load varied between 36 and 41, tending to stay near the mean.

#N canvas 0 22 454 304 10;
#X obj 226 85 inlet~;
#X obj 226 109 bonk~;
#X obj 226 146 outlet;
#X obj 303 110 fiddle~;
#X obj 305 82 inlet~;
#X obj 298 141 outlet;
#X connect 0 0 1 0;
#X connect 1 0 2 0;
#X connect 3 0 5 0;
#X connect 4 0 3 0;

i wonder if these results hold for more complicated patches.

never mind - the answer is here: http://puredata.hurleur.com/sujet-145-tr808-sample-player

pool looks really awesome but i think that coll might be better suited to what i need. thanks!

in the past couple days i have built a very extensive saving system that didn't work, and now i'm back to the drawing board. it's okay, because i looked at the documentation for qlist and textfile in max/msp. you really do get what you pay for. i'm now working on using qlist to assign each drum sound a dedicated line in the file, and then have qlist navigate to the correct line and make changes. this seems necessary to prevent the creation of a new line each time a change is made, which is what happened with my first idea.

my questions are:
is there a way to tell qlist (or textfile) which line to go to? i know that you can in max but those commands don't work in pd. maybe a port of max's qlist and textfile objects is in order.

is there a way to automate repetition of messages? i could navigate to the correct line by sending "next 1" to qlist x times where x=the number of the line i'm trying to navigate to.

are there any other objects (including third-party) that can manage textfiles similarly?

thanks, leo

thank you! this works really well for what i need. now to adapt it...

i don't know, that patch seems a bit too overqualified for what i want. i am just trying to make a drum machine that plays pre-recorded drum samples. some of the samples are very long, and if, for example, i want to play a cymbal on two consecutive beats, i can't just tell a tabread4~ to start reading again on the next beat because it will cut off the first.

short of creating a different tabread4~ for use with each sound on each beat (8 sounds * 16 beats = 128 tabread4~s), is there a way to do this?

leo