I like the nature sounds idea a lot. Our hearing is immensely sophisticated and we could probably develop an immersive experience of this for network monitoring with the right metrics. The idea of a secops analyst sitting by a virtual reality campfire, listening to their "surroundings," and then investigating based on environmental changes is pretty cyberpunk, and absolutely doable with something like this. Nature sounds as a complete UX metaphor is pretty amazing, and rich sound environments may be the missing immersive bit in VR/AR games.
This is timely, as just last night I was by a fire listening to a field of crickets and wondering if I could use a GAN to learn to interact with their fabric of noise and then convince the crickets to modulate their chirps according to the GAN's feedback. I then wondered how plausible it could be that this had already happened to us. Indeed, that kind of night. But these are the most interesting problems, and adversarial AI for insect control using their own audible telemetry has to be the most viable path for having it interact with biological systems. If it can't pass an insect Turing test, it probably can't pass a human one.
Thank you for posting this, it was truly a hidden gem.
I have heard that humans are much better at discerning small differences in frequency and recognizing patterns in audio than we are if we look at a graph/chart/table visually. Virginia Tech and others have been doing some "data sonification" work in recent years that seems really interesting.
Destin of the Youtube channel Smarter Every Day did something similar to your hypothetical cricket experiment with fireflies and LEDs recently where he was able to alter the group's behavior.
Wow. So, we would have a control input for the model for correlating chirps, where mics and thermometers around a field could measure chirp frequency against a baseline of temperature. Next question would be whether crickets migrate toward the chirp-signaled higher temperatures, and whether they submit to signalled chirps or trust their own in their migration decisions. Sound is just relatively low frequency data, and a field of crickets would give a uniquely large sample signal data set.
Fun! I didn't know somebody had an OSS project that did this - 22 years ago.
The abstract of their paper reminds me of the story about the early Amazon team setting up a bell to ring in their office every time they sold a book.
Two years ago, I wrote a similar tool [1] to play notes in a C-Major scale based on the moving average of network events on a network interface.
The amount of dependencies the project has makes the cost of getting the thing to run much higher than the satisfaction of it running! Not fun!
I will go out on a limb and say that a times-series of system performance somewhere on your desktop is a better solution long term than listening to sounds that you play continuously from a speaker.
The Solaris `snoop` command for capturing network packets has supported a -a option for at least 20 years which listens to packets on /dev/audio and comes with a warning that it "may be noisy". It used to be fun to run on a Sparcstation!
When my employer had a lot of SunOS 4 SPARCstations on coax Ethernet, we had a specialized appliance for this. It was based on a re-badged black Toshiba laptop with an orange display (T3100, maybe). IIRC, traffic sounded like a geiger counter.
(Hauling out this device would've been even cooler, had we dimmed the cube farm overhead lights, and worn 37337 hax0r sunglasses.)
Nice, having sound in the loop for human-computer interaction I think is way undervalued. We have notification sounds yes, but would love to have more background tasks with some kind of subtle audio signal.
Especially now that these apple silicon macbooks turned so quiet, before you could still sort of get a hint that your computer is acting strangely when all of a sudden the fan would speed up.
Human-machine interaction in general. Humans are incredibly attuned to subtle unexpected changes in routine noises, and "that didn't sound right" is usually the first sign that any equipment is starting to misbehave.
Just the other day on here someone was discussing their fan speed modulation software they ran on their laptop, based on CPU usage on remote VMs or something similar.
I used to run this on my home network. It was pleasant and occasionally useful. The code was pretty unstable though, and the project looks totally unmaintained now. E.g., the “news” link redirects to a vacant domain. Sourceforge complains about the code being in CVS and suggests migrating to Subversion.
Perhaps some HN visibility will encourage someone to adopt and resurrect it.
I can't find a bookmark I had which was an online platform for this - but which would integrate various APIs you connected, or maybe it was just for a website's traffic - where you could set a certain noise to trigger, say on new visitor or when a certain page like signup was touched.
I thought it was a nice passive way to get notifications, and something that could be motivating depending on what you have it linked with.
It was called choir.io. It allowed you to translate streams of events into sounds. You would categorise your events into a matrix of good-neutral-bad with different levels of severity, and point them at choir.io. The service would then play samples that fit the category in a web client. There were different sound themes as well.
They probably couldn't make it pay, and it disappeared, but I sorely miss it. Would have been nice if it was open sourced.
This is timely, as just last night I was by a fire listening to a field of crickets and wondering if I could use a GAN to learn to interact with their fabric of noise and then convince the crickets to modulate their chirps according to the GAN's feedback. I then wondered how plausible it could be that this had already happened to us. Indeed, that kind of night. But these are the most interesting problems, and adversarial AI for insect control using their own audible telemetry has to be the most viable path for having it interact with biological systems. If it can't pass an insect Turing test, it probably can't pass a human one.
Thank you for posting this, it was truly a hidden gem.