Hi,
There seems to be a discrepancy in corpus loading between the various provided catart mubu examples. Made a screen recording that shows the difference between a few example patches.
I have uninstalled and re-installed both packages, issue persists.
Running :
Max 9.02
catart_mubu 1.6.0
Mubu For Max 1.10.11
Thank you for any pointers how I could fix this.
Best,
Sergio
Hi, the differences you observe are due to different default settings for the segmentation mode: regular chop for the intro patches, and more musically useful onseg onset segmentation for others.
Let’s think about what would be the best default settings and clearest documentation of these choices.
Cheers
Hi Diemo,
Thank you very much for your swift reply. Made a new screen recording where I change the segmentation mode as sugestted to regular chop in the simple example patch, however the discrepancy persists. I might also note that this behaviour is only happening in Max 9, in Max 8 they all load the same.
Any insight most welcome, thank you.
Best
Thanks for trying, now you found a little error: the chop subpatcher (double click on camu.anaysis to open it) seems to have been inadvertently saved with a chop.size of 0, meaning each sound file is used as a single segment!
Thank you Diemo !
Getting closer, although even when using the exact same values in the intro patch’s “descriptor-analysis” and in the simple example’s “camu.analysis corpus” the resulting analyses’ scatterplot is slightly different, mainly noticable in the low end region, is this expected behaviour ?
Many thanks & sorry to be such a nitpicker 
There are 4 more settings that diverge, here are some good values: @descr.winsize 2048 @descr.hopsize 256 @descr.minfreq 0 @descr.threshold 0.25
I appreciate your attention to detail!
Yes !
Thank you Diemo.
Im achieving consistent results now across all example patches, great.
Thank you for your persistence.