ok I’m trying it on the same dataset, and here are my comments, verbose, so you can comment back!
- first impression:
- nice looking
- good simple (cliché) demo
- clear interface upon making a new corpus (I skipped the tutorial altogether to see how ‘intuitive’ it was)
So I decided to make a direct comparison with the modular dataset we used for @groma’s Fluid Corpus Map NIME paper
- irritant - processing 0 files, initializing, does not move at all… caps at 33% cpu, so might be stalled for real. I quit. after restart, it does behave. Their analysis feels slower than FluidCorpusMap but i’ll need to retry. It is definitely slower than CataRT.
- bug: the window zoom is not zooming the space but zooms the click values… reverting to original size helps a bit - clicking on fullscreen make it work (yet does not zoom the space)
- irritant: the sequencer is rolling all the time
- cool: the colour scheme for the interactive clustering with one slider
- sad: I really miss the other drawings I get from other reductions - a single timbral space like this might or might not be inspiring.
So my conclusions: it works relatively well, it is closed enough to be beginner-proof and might entice people to code their own after playing with it and being irritated by someone else’s interface 