jit.peek~ timing precision on image sonification


    Mar 23 2021 | 7:26 pm
    Hi all! I'm finding an annoying precision fault on an image sonification patch i'm working on, and i don't know how to solve it. I guess it is related with the way jit.peek works, or maybe with the signal feeding it, but i don't encounter a proper solution for it, that's why i write to the forum to see if someone can reproduce the problem and guess what's happening.
    When sonifying an image like the one i'm attaching here, where the visual elements are formed by periodically separated vertical lines to create steady rhythms, it sounds like there are pixels missing and timing errors. You can load this image on the patch i attach and hear for yourselves. The sonification kernel of the patch is inside the "p sonifica" subpatch, where -i guess- should be the problem located....
    Does anyone have a clue on what's happening there? how can i solve this and get a precise timing on this sonification method?
    thanks!!!

    • Mar 24 2021 | 10:28 am
      I'm attaching an image and code of the sonification bit, if that helps figuring out better ways of doing this. Basically, jit.peek~ is reading pixel data from a jit.matrix, and its contents being scanned column by column. Each of the 1050 pixels in a column are transformed into magnitude values for an ioscbank, where the lower pixels represent lower pitches while higher pixels represent higher pitches.
      Although the whole construction works reasonably well, I'd like to improve sound quality. The main improvement i'd like to achieve is related to timing precision as i mentioned on the previous post, but also achieving harder transients (in order to have stronger percussive elements).
      Maybe the fact that i'm not feeding ioscbank with phase data is affecting somehow the sound quality, but i don't know what kind of phase data should i feed the ioscbank with, as i am working with generated images...
      Any hints on this would from the jitter-msp wizards would be very appreciated, Thanx!
    • Mar 25 2021 | 4:18 pm
      I there! I think that if you use very low freqsmooth and magsmooth values in your ioscbank~ you will get harder transients.