Recently I completed my live performance tool Instant Decomposer. It is done in Pure Data, some new externals were specially written for the project. An introduction video clip is on

The project is documented on http://www.katjaas.nl/instantdecomposer/instantdecomposer.html

Katja

That is way too cool!

yes, please upload.

Katya, thanks for sharing. I learnt a bit more about creating externals too thanks to you. I compiled them and already having fun!

S

OSX binaries

Attached is a .zip with binaries slicerec~.pd_darwin, sliceplay~.pd_darwin for OSX and a patch slicerec~-help.pd.

The source is on http://www.katjaas.nl/beatdetection/beatdetection.html

If something about these objects is unclear, please let me know. And if you find an interesting way of using the objects, I would be happy if you want to share that.

Katja.

http://www.pdpatchrepo.info/hurleur/slicerec&playOSX.zip

Youtube performance is great, congrats!
Great documentation on your site too, keep going!

AZ

nice work!

Hey Alberto, you are the one who wrote that very useful MinGW-for-Pd-externals tutorial! Indispensable tutorial for me when I compiled externals on a friend's windows system. Thank you for that.

Katja

Great to know that it was useful...

Thumbs up for instant decomposer!

Alberto

dear mad audio lady
don't listen to your dad
he has no clue at all
thou is now one of my audio heroes

i want one instant decomposer too!
but it seems that i'm to stupid to even find the files...

by the way there are some different image objects in extended

*** farout no doubt ***

great documentation also

Yeah, where are those Instant Decomposer files.....

I'm workingt to make at least bits and pieces available for everyone to use, but it's a complicated project. Some externals, amongst which slicerec~ and sliceplay~, were written for I.D. because the process is not precise enough otherwise, or too CPU-intensive. They all run fine on OSX 10.5. But last weekend when building the classes with latest Kubuntu on a friend's computer, we found (some of) the objects to fail. So next time is for debug and test, and for windows builds.

Part of the set will be [soundtouch~] realtime pitch shifter with Olli Parviainen's SoundTouch library built in.

When build&test is done I will do a couple of demo patches in Instant Decomposer style and upload things to appropriate places on this forum. You know what, if there's some new I.D. stuff to share I will leave a message here.

Katja.

BTW I ran Instant Decomposer on a three year old Asus EEE netbook with Ubuntu. Ok it did 100% CPU load but it was fully functional. Now working on Mini Decomposer... for imaginary touchscreen tablet, yippie

yeay!
sounds great and really worth the wait

@katjav said:

OSX binaries

Attached is a .zip with binaries slicerec~.pd_darwin, sliceplay~.pd_darwin for OSX and a patch slicerec~-help.pd.

The source is on http://www.katjaas.nl/beatdetection/beatdetection.html

Katja.

BTW, attached you can find the compiled .dll for windows in a zip file.
Seems to work ok with the help patch.

Alberto

http://www.pdpatchrepo.info/hurleur/slicerec_play09.zip

binaries and GUI module, half an Instant Decomposer

I have posted [slicerec~] and [sliceplay~] binaries for OSX, Linux 32/64 bit and Windows on the extra~ page. Included is an algorithmic trigger module [slicycle~] with GUI, like I use it in Instant Decomposer.

Katja

p.s. most recent versions are now sliceplay~ 0.9.11 and slicerec~ 0.9.1. The 0.9 versions where not 64 bit ready. In sliceplay~ a bug with playback speed is fixed.

Wow. Thank you for being awesome.

Excellent!

Thank you, and keep well ~ Shankar

Wow!

Brilliant!

This has made my day!

All the Best

John

Perfect - care to elaborate on the generation of the parametric values? General method, theories of control, etc.

Perhaps some frequency content detection in order to place sounds rhythmically in a more coherent fashion? An alternative being specifying what "kind" of sound it is before the program takes over.

You make the imagination expand. I have debated giving more control to the computer and now you have made the computer seem worthy.

+1 for the pulse BPM.

Ponzo, note generation in Instant Decomposer is deceptively simple: beats, 8th notes and 16th notes are all generated, with random selection of slices and playback speed. Some ticks are skipped, when a slice is still playing and not allowed to be interrupted. Therefore, with long slices you don't hear the pulse.

In the meantime I have developed a kind of sequencer-method using mathematical patterns which can replace random slice selection. So, in addition to pulse, there is pattern structure. This method is also built into Slice//Jockey, a shared patch which you can find from this forum.

But pulse and pattern are not inherently musical operators, and beautiful sounds do not automatically make beautiful music. Should a sound automatically get it's niche where it can blossom, routed by better analysis in the program? No, I want my computer to only do the stupid hard work, like cutting samples and counting beats. It must leave esthetic decisions to me. So, at this point in my very immature research, I am looking for ways to regain control. A better interactive GUI could help with this. A GEM GUI, possibly...

Katja

Edit: got it loaded up and it's just very pleasant to work with. The CPU load is not present...excellent externals those are!

I'm afraid that I have not had a chance to work with the Slice/Jockey yet but I am looking forward to it. I would love to see the pattern interface.

I suppose that if beats are generated separately from subdivisions that explains the sense of pulse.

I agree that the "stupid hard work" should be left to the computer while the user can influence the feeling and add the human element to the mix. I am always trying to think of ways to pass along formal structure to the computer and let it decide how to make this or that sound like a phrase or arrival point. Once you have it done, you feel like you are within a dark maze of form - the phrase ends before you're ready because you told the computer it was 32 bars and time is up!

I would like to take the time with your patch to let it record a percussive motive into its separate slices, store the time between each attack, and then use those values in a markov chain to recreate motive-derived variations. Is that relinquishing control or regaining it? I understand an argument for each.

I believe that many traditional performers may find this sort of manipulation more satisfying - a soloist, for example, may play an ostinato that your program repeats for them, but not just as a loop (strict ostinato), an idea. Speed, density, variation, can be altered...though that introduces many more control issues for a performer to worry about.

J.P.