UnUnUnium

Hello everybody. I am trying to create a live sampling patch for a course I am on which interacts with a hand percussionist, and was hoping for some help.

I am using various objects to gather spectral aspects of the sound- for now I am keeping it relatively simple and using centroid~, segment~, and maybe noisiness~ to give me the spectral average, attack count, and aperiodicity respectively.

I am attempting to gather this coninuous stream of data and pack it into a lists of exactly 128 steps of length, each of which I will scale into a value between 0-8.

The reason I am doing this is to have step sequence patterns which correlate to a sort of histogram of various aspects of performance- time between attacks, volume, etc, which then playback in response to the player.

I am easily able to collate a list of 128 values using zl group, scale them to 0-8 using a list scale object, which feeds a multislider, which feeds a sequencer...

However, what I really need is something that will take a history of data that is an arbitrary length, say a list of 582 values for example- and 'quantize' them into a list that is 128 values long and between 0-8 values high that resembles the larger list, albeit more 'stepped', and I need to be able to do it with a "bang", when the current playing sequence ends.

Any guidance, magic object suggestions, vague hints, or different approaches would be greatly appreciated.

 My current hang up is this:
   I am using various objects to gather spectral aspects of the sound- for now I am keeping it relatively simple and using centroid~, segment~, and maybe noisiness~ to give me the spectral average, attack count, and aperiodicity respectively.

  I am attempting to gather this coninuous stream of data and pack it into a lists of exactly 128 steps of length, each of which I will scale into a value between 0-8.

   The reason I am doing this is to have step sequence patterns which correlate to a sort of histogram of various aspects of performance- time between attacks, volume, etc, which then playback in response to the player.

   I am easily able to collate a list of 128 values using zl group, scale them to 0-8 using a list scale object, which feeds a multislider, which feeds a sequencer...

 However, what I really need is something that will take a history of data that is an arbitrary length, say a list of 582 values for example- and 'quantize' them into a list that is 128 values long and between 0-8 values high that resembles the larger list, albeit more 'stepped', and I need to be able to  do it with a "bang", when the current playing sequence ends.

  Any guidance, magic object suggestions, vague hints, or different approaches would be greatly appreciated.

listogram-help-with-data-management

> Any guidance, magic object suggestions, vague hints, or different 

> approaches would be greatly appreciated.

Did you have a look at histo and table objects?

>  Any guidance, magic object suggestions, vague hints, or different 
> approaches would be greatly appreciated.

Did you have a look at histo and table objects?
Or, if you have jitter: jit.histogram

Thnak you for your reply. Yes, I've looked at table- I am pretty much using multi slider in the same way- I've looked at histogram, but I can't see how it might solve my problem, unless I'm missing something.

I need to take running tab of incoming values of an arbitrary length, and squash them down into a list of 128 values at any given moment, when a bang is recieved...That list need to be a quantized representation of the longer list.... I have figured out how to scale the values vertically (using mxj ej.lscale), but I can't figure out how to do it horizontally- that is the problem....

Thnak you for your reply. Yes, I've looked at table- I am pretty much using multi slider in the same way- I've looked at histogram, but I can't see how it might solve my problem, unless I'm missing something.
  I need to take running tab of incoming values of an arbitrary length, and squash them down into a list of 128 values at any given moment, when a bang is recieved...That list need to be a quantized representation of the longer list.... I have figured out how to scale the values vertically (using mxj ej.lscale), but I can't figure out how to do it horizontally- that is the problem....

First step would be storing the arbitrarily long list, then, when a bang is received, use [zl len] to get the length, and then divide that number by the length of the list you want to end up with.

- say you're list was 356 members long, you would get 356/128 = roughly 2.78. Round that up to nearest int, and split the list into chunks of that size (3), run it through [mean] and pack it back down as a list (uzi?)

What you end up with is a patch that quantizes and averages at the same time, spewing out close-enough looking 128-long lists.

This is of course of the top of my head, so there could be stuff missing, but that's the way I would go, at least.

First step would be storing the arbitrarily long list, then, when a bang is received, use [zl len] to get the length, and then divide that number by the length of the list you want to end up with. 

This is of course of the top of my head, so there could be stuff missing, but that's the way I would go, at least.


Excellent! That is the ticket. That is exactly what I want- a quantized semblance of the shape over time.

This solution should work perfectly for what I want it to do. I will experiment with zl length and perhaps slice and see what I come up with..I have a feeling I can pack it all back using zl group....

Excellent! That is the ticket. That is exactly what I want- a quantized semblance of the shape over time.
  This solution should work perfectly for what I want it to do. I will experiment with zl length and perhaps slice and see what I come up with..I have a feeling I can pack it all back using zl group....
  Thank you very much for the insight!
Cheers, Jeremy


Listogram? (help with data management)