Sunday, July 26, 2026

[MusicViz] Spectrochrome Update - Pitches Detected Pane

Had a really productive day and a half on my "Spectrochrome" project again after a 2 week hiatus!

(Apart from running out of steam, I was probably starting to stretch a bit too much on top of still having my day job and/or trying to get some "non-computer-based relaxing/life-living time" in...  And with no external pressures on what should and is largely still an actual fun hobby project, a break was indeed in order!)

Anyway, I finally got back to doing some "new" stuff again, after getting really worn down trying to get that massive colormaps refactor MVP stage sufficiently complete that I could shove it out the door, and not have to deal with the rest of the annoying + HARD problems requiring further intricate refactoring work for a while. Some of the highlights were:


* 1) Got a new "About" / Help-Text panel set up

(Don't read too much into this text in the screenshot just yet! :P  It's still really just a roughly thrown together wad of fluff to have at least *something* - which I've since begun the interminable process of gradual refinement on...)
 

* 2) Also bashed together some stuff to show a floating panel containing details of all the detected notes and their intensities... 

(This was one of the big outstanding features on my roadmap / todo-list, as I needed a faster + more interactive way to see this data, without having to first export it to a CSV file that I then needed to bring into some other tool to read..)

Disclaimer:  This screenshot shows the current "incomplete" state of this - hence why all that empty space to the right of the table is very much needed + justified. Specifically, there's a bunch more work to come soon to show what other contenders for the "strongest pitch" at that moment may be!  (This was actually the way I had the very first version working, but very soon reduced down to just the "strongest pitch", to try and bring that feature under control.  Anyway, hopefully getting this working may reveal some interesting insights!)

Another direction to take this is to try plotting these "strongest pitches" on/near the scrubber while doing the recording

   i.e. There's currently a problem with the generated videos where you can't really get a good feel for the notes that are getting hit / emitted when the scrubber hits them. I feel like we really need some visual feedback that is seen for a few frames before the sound hits and the line actually crosses the point in question to really "sell" the mapping between sounds and visualisation...  (I look forward to that, and/or trying to draw some sort of trace showing how we think the melody might be flowing between these points!

 

* 3) Finally, have managed to do a bit more research into how to try to graft in some "Equal Loudness Curve" support so that I can adjust the intensity that various frequencies are being visualised + processed at - Hopefully when working, this can help dampen out that pesky band between "C7 - C8" that tends to fill into a solid block of bright colour, as it seems all the other harmonics all merge/intersect + resonate at that frequency range... (along with compensating for fundmentals actually rooted at C3/C4 range always ending up triggering a detected peak an octave higher)

(Failing that / or in addition, I really should adjust the colormap to make that range more desaturated so they are play less of a role)

No comments:

Post a Comment