I’m pleased to say that the Technocolour Dreamcoat* performed amazingly at London Decom. I got a whole 7 hours out of the batteries, running a combination of sparse and power-hungry patterns throughout the night depending on my whim. Heaps of great reactions and maybe a dozen people asking about how it was built. I ended up writing “FastLED” on one girl’s arm, please say hi if you found us
It was damned hot in there though, and it’s one thing to be wearing a great big fluffy coat but another one entirely if that fluffy coat generates twenty watts all by itself! So maybe my future 200W version might be kept in reserve for open air environments :-). But as it was I didn’t quite sweat enough to short anything. Apparently.
One thing that didn’t work so well as I’d hoped was the sound reactivity. I found that the room was just so noisy it was difficult to set the threshold such that it would only “trip” on a bass beat, it ended up being all-or-nothing mostly. I’ll start to look into HFTs and the MSEQ7 to try and fix that I guess.
Again, huge thanks to everyone here for helping me out along the way. See you at Dutch Decom next!
(* The name of the project is most definitely “Technocolour Dreamcoat”, not “Disco Sheep”, as one of my friends suggested
Sound’s a tough thing, however I’m fairly happy with my recent foray away from MSGEQ7 and into FHT by OpenMusicLabs. It was a bit tricky to get going but the results are pretty cool, even with a simple routine.
Any chance you could make a recording, with both the music and background noise during a future a show? Improved analysis for the Teensy Audio Library, including code to automatically adjust gain and thresholds, is something I really want to work on. Some good recordings of music+noise from actual shows would be incredibly helpful.
I could try and do that, though I don’t have any kind of flash storage (SD card) hooked up. I wasn’t actually using the Teensy Audio lib yet; and I think part of my problem was I was only sampling from the mic once every tick of my loop() method (every 1/60 second.) This worked reasonably well in the quiet environment of my living room, but not so much in a crowded club.
Yes, I would very much like to gather a small library of noisy recordings from mics on wearable electronics. They should be exactly the same 16 bit, 44.1 kHz mono stream the Teensy Audio Library would get as input.
Already, some people are working on much more sophisticated musical analysis like note onset detection and beat estimation. The trouble is these algorithms are a lot easier with a pristine original file than with a cheap mic that’s positioned poorly, listening in a club or event with people talking and background noise and reverberation/reflections, and pickup of sounds like ruffling of the clothing it’s mounted upon.
It’s pretty much impossible to develop electronics and software in those environments. The only hope for ever getting a really awesome library that “just works” in those extremely poor conditions is collecting a large set of terrible audio for testing.
