A while back I worked for a team that had wired up Jenkins to a speaker in the office. Each type of event would trigger a different Zelda sound effect. Victory music for successful deployments, game over music for breaking the build, etc. Notably, dev server exceptions were connected to sword clashing sounds.
It led to debugging situations like this:
"Okay, so that's two clanks when we click on this button, but if we do it on this other page it's only one clank. Hm."
Turns out, there are some bugs that are easier to detect this way. Looking at timestamps doesn't really give a "sense" of whether two events are in a tight causal link. With sound, you can immediately "hear" that two adverse events are occuring at identical intervals every time. What's great is that the sound information doesn't take up any additional attention. It just fades into the background if you don't need it. When there is a pattern, it stands out.
At a previous gig, we had a lot of systems named after animals, and I'd joked about wiring our alerting system up to emit the animal sound for the system that was paging, but I think I'm actually going to do that at my next gig. Outside the land of software, engineers debug systems using sound all the time - I think we're underutilizing our senses and our pattern recognition.
I always wonder at the madness of elevator design. Most of them "ping" as they pass a floor for accessibility. But that means the blind rider must first know the floor he is on, then count the pings to the right floor.
It should simply announce each floor as it passes. I know from experience that you can make quite comprehensible voice sounds by connecting a 5V I/O pin to a speaker, so there is no excuse. (All you do is digitize the speech, then turn the pin on and off as the wave crosses 0.)
Ditto for every device that comes with a chart saying which LED blink pattern means what status. Like the "fast blink" and "slow blink". The LED is blinking, is that "fast" or "slow"?
I've never been in an elevator that announced the floors with anything better than a ping. This is despite elevators being controlled by single board computers since around 1980.
The buttons have had braille on them for a long time.
Interesting. OTOH, while old builds here (electromechanical, no microcontrollers, <1990) lack even the basic ping, I haven't been in an elevator built in the new millenium which wasn't excessively chatty. (Central Europe mostly, but the same experience everywhere I went: "ground floor, main entrance, access to train station; mind the door; door is opening")
Interesting you say that, literally every lift in a public space that does any sound at all around here announces each floor with proper spoken speech, not just a ping.
I wonder if it is due to some ancient regulation mandating rings/pings that nobody bothered to update for current tech. When i lived in Warsaw (Poland) for a while, every single elevator i came across - including those in very old buildings - had voice recordings for announcing the current floor. I almost learned polish numbers from them :-P
In a lot of modern buildings in Hong Kong, lifts announce floors in three languages: English, Cantonese, and Mandarin.
I'd imagine in Macau, they might use four (those three plus Portuguese); the public transit stop announcements do, and quite frankly I'm not sure how they guarantee that they can complete the announcements before the next stop.
the main reason why elevators in Hong Kong do not is that most of them are high speed high-rise elevators and the numbers just whip by. It would probably be very frantic sounding to use Morse if you went from ground to 50 in less than a minute.
Instead they only announces floors that are stopped on, so this works fine.
One benefit of using spoken vernaculars rather than Morse is that Morse has to be taught; Hong Kong historically draws large portions of its population from migrants originating from all over China, where until recently there was wide variance in schooling and literacy rates.
I know of at least a few in London office buildings that speak. Especially buildings with large banks of elevators that tends to have "fancier" control systems that manage all of the elevators.
Canadian here - even my condo building elevator announces floor numbers with a (pretty well-digitized) voice. And believe me, my building is no marvel of modernity.
A fun demonstration of getting comprehensible sound via one bit sampling is that you can use a tight loop on a C64 to read from any number of input pins (such as the tape input) to sample sound, and play it back either by writing to any number of outputs and connect it to a speaker, or indirectly by using the signal to toggle the sound chip volume up and down.
So not only can you get comprehensible voices from a 5V I/O pin to a speaker, pretty much any CPU newer than the mid 70's will be powerful enough to drive it.
So I agree - there's not really any excuse other than that people haven't thought about the UI.
Careful not to burn anything out... Though the C64 is pretty indestructible - I connected all kinds of things to the IO lines of it back in the day... And a few inadvisable cases of soldering things straight to the user port pins (I'm amazed I never destroyed any machines.... that way)
A tip is to turn off the screen during sampling and playback, as otherwise the video chip steals a lot of memory cycles.
You might then also be interested in this far more impressive playback with the C64:
The one edge case is that the bells work for all blind folks, not just those who understand the language spoken by and announcement. Of course, a blond person in that case could just count announcements …
> Of course, a blond person in that case could just count announcements …
Exactly, he's not worse off.
Besides, I've traveled in foreign countries where I literally do not know a single word in that language. You start picking up words by association almost immediately.
The "ping for accessibility" is a holdover from ancient times (think 1910s), when there was an actual bell that was rung by the elevator's passage (one ding up, two dings down, IIRC). Current technology has (and uses) far better capabilities, either by (common) samples or by (uncommon) speech synthesis: "ding" "Floor 6." "Doors opening..."
I have noticed that some street walk/don't walk signs now have a voice rather than just a tone. It sure took a long time. Though the voice would be improved by saying "walk east" rather than just "walk".
-At a former employer, I had coded some embedded devices to spit out their bus address and current error code in morse code using the buzzer in the keypad. Saved me lots of time during initial debugging of new installations, then I'd just disable the buzzer before heading for home.
(At least) on one occasion I forgot; years later a puzzled chief engineer calls me and says that his first officer had told him that E209 had a pending IGBT failure while they were discussing issues with the equipment we'd delivered over the din of the buzzers. Would that hypothesis have any possible merit?
It might depend on the person, but sound can be a powerful way to perceive the world. When I was living in another country and learning the language, I found that the difference between a lot of bad or mediocre language learners and those that were really good, was that those that were good eventually stopped thinking in terms of grammar rules and dictionary words and started remembering phrases by how they sounded.
If you say something that is grammatically incorrect in your own language, most native speakers will tell you it just "sounds" wrong. They can't even explain the rules of why it is wrong, but they know it sounds wrong.
This reminds me of when I was helping a friend debug his home-built computer. Couldn't get a POST message out of it indicating why it wouldn't boot. After a while of searching, discovered it was because there wasn't a speaker connected to the non-existent pins for the speaker (there were no pins, just the soldering points for them). Once discovering that, used a multimeter to read the POST error - no GPU detected. Fun times...
Once I was introduced to an oscilloscope, I much preferred that. I bought a nice one off of ebay for about $60, and another peach of a scope from the pawn shop for $40.
That's super interesting. I would imagine another benefit of that system is pattern detection. It would be much easier to notice a "beat" happening than with log messages, where you need to put things on a timeseries representation to see those patterns show up.
It led to debugging situations like this:
"Okay, so that's two clanks when we click on this button, but if we do it on this other page it's only one clank. Hm."
Turns out, there are some bugs that are easier to detect this way. Looking at timestamps doesn't really give a "sense" of whether two events are in a tight causal link. With sound, you can immediately "hear" that two adverse events are occuring at identical intervals every time. What's great is that the sound information doesn't take up any additional attention. It just fades into the background if you don't need it. When there is a pattern, it stands out.