I started using flucoma recently (in Max) and have run into some issues when segmenting and analyzing files. I’m having extremely long wait times when analyzing files using fluid.bufonsetslice~, fluid.bufspectralshape~ or fluid.bufmfcc~. There is typically little wait time when segmenting, and I’m getting a pretty clean segmentation. I’m attempting to analyze a 2:33 wav file, which is resulting in wait times up to ~5 minutes. For comparison, the same sound file can be segmented and analyzed in CatArt in ~30 seconds. I would love to use flucoma but having to wait twice the length of a corpus any time it’s loaded feels very cumbersome (especially with such a small corpus). Is there anything I should look at when analyzing segments to speed this up?
Hi @drewmfarrar ,
That does seem strangely long. I think we’ll need more info to troubleshoot. Can you provide your patch and/or sounds?
1 Like
Hi @tedmoore here’s a drive link with the max patch and the audio I’m using:
https://drive.google.com/drive/folders/1v4W3MHoLIOwFpDI-V65NPo-X-2wQyhdw?usp=sharing
Thanks for your help!
I think the problem is your trigger object after the zl stream: it has t b i which means that the numframes you’re passing to the processor aren’t what you want: they’re big negative numbers, which probably means that the processor is just doing the whole buffer every time.
Try changing to t b l to pass on the list from zl stream instead.
3 Likes
That fixed it! Thanks for the help!
1 Like