< Back to IRCAM Forum

OmQuantify Stack Overflow

OpenMusic
#1

Dear OM community,

I have been stuck recently because I am trying to “omquantify” a fairly large “chord-seq” (lots of pitches, ca. 10 minutes in duration).
The computation is too large and I hit stack overflow even after increasing memory to the maximum of 261120.

I can’t make the “chord-seq” smaller because of the process I am using , however, is there a way to partition or extract it in segments so that I can quantify it?

Hopefully there is an easy way to do this, but any feedback is very welcome!

Thank you
eb

#2

Hi,
this seems very odd. Can you at least send the chord-seq, or patch to check this out ?
It could be related to something else.

K

#3

Hello, you can also try to segment your chord-seq in the “Segmentation/Analysis” mode of the chord-seq editor.
http://repmus.ircam.fr/openmusic/dev-resources/analysis#the-segmentation-mode

This is not documented in this page but you can also select “KANT-SEQ” as a “New Analysis”, then make the segments by selecting chords and pressing “s”.
Double-clicking on the segments will allow to set omquantify parameters for each one of them if needed, and “Analyze segments” will compute the rhythm trees.

Then you can export the quantified voice with ‘concatenate-kant-voices’

Screen-Shot-2016-12-13-at-12.25.38.png

#4

I’m working on a tutorial about it ! Notify that the actual version doesn’t recognize silences inside a segment, only the silences finishing the segment

thanks to Karim, there’s an alternative (see attached) but it’s not perfect (only one tempo, for example)

I hope we could expect an integration of TRUE-DURATIONS algorithm inside the Kant analysis.

……………………………………………………………………………………………… jerome

kant_segmentation.omp (24.1 KB)

#5

Here’s a closer analysis with the same measure metric and the same parameters for the two VOICE objects.

@Jean, have you seen in the “kant message” that the last measure is (3 4) and not (4 4) ? It’s because, at last measure you don’t fill 4 times !

As I’m studying the Segmentation and analysis Framework in general, I would like to know if you will improve it… It dates from to 6.6 version…

Cheers~

……………………………………………………………… jerome

kant_segmentation1.omp (24.3 KB)

#6

Hi Jerome

The quantification at work here is the omquantify function, so I think it behaves the same.

I have no specific plan for more work on this in the short term, but I can fix bugs and issues.

Jean

#7

hello Jean, and all,

As KANT-SEQ and TRUE DURATIONS have many things in common, do you (or any people who can code for OM) think you can fusion these two process, in order to have KANT PARAMS with « real » tree (negative values for silences) ?

Cheers~

…………………………………………………………… jerome

#8

Using the following patch the “analyse-segment” of kant-seg will use true-duration to get the durations of the segments.
Let me know if it works. If it does, I can include it in the next OM update.

kant-seg-analyse-one-segment.lisp (949 Bytes)

#9

Salut Jean,

As my Mac is out of order, I’ve made a test on a PC. It seems to be ok but each time I launch an analysis, I loose the last note of the CHORD-SEQ.

Another little thing : sometimes when I select notes and type “s” I have this message : I can’t build an chord with these parameters

Well, Jean, it’s almost perfect !! :slight_smile:

… jerome

#10

Erratum…

Actually, it works wery well !

THX so much…

… jerome

#11

The kant-seg worked and I was able to make progress thanks to it!
Thanks so much for the help

Best
eb

PS: Sorry Karim, but I couldn’t attach the chord-seq for some complicated reasons :slight_smile: All good now