The idea is that instead of just being limited to encoding numbers into images only, lets encode numbers into sounds aswell.
Let xxx-yyy-zzzzz be a random sequence of numbers, for instance 000-000-00000, where xxx-yyy is the normal 2 chunked major system images. In this case 000 could be “sauce” which in my mind would be the youtuber VSauce. zzzzz is the sound that gets created when xxx interacts with yyy. So in this case where peg1 = Vsauce and peg2 = Vsauce aswell, we could picture Vsauce slapping a cloned version of himself, where the slap sound is zzzzz. I’ve already tried coming up with some potential categories for every z: (instrument, ocatave, note, note, note), but this is where I lend the shovel to this forum if so to say…
One thing to remember is that if you’re trying to create a compound sound where each digit is some kind of attribute of that sound (like instrument, pitch, actual note etc) then you’re only getting the data compression efficiency of a 1 to 1 conversion. You have to actively consider every digit and accurately encode each digits attributes into the complexity of the compound sound.
You may be better off limiting the sound encoding to two digits. Its much easier to come up with 100 unique singular sounds and more efficient in actual scene building than it is to have to accurately read and encode 5 different digits singularly.
This way you get 8 digits per scene with only three intentionally encoded elements, versus 11 digits with 7 intentional elements. A much better data compression and simpler scenes to keep track of.
You could do 10 categories and try to find 10 unique sounds that fit each.
So say that 0 is rock instruments. You could have a clean electric guitar, a distorted metal guitar. A drum roll, a snare pop, a crash cymbal, slap bass, hammond organ chords, proggy rush keyboard solo sound, a wailing vocal like Axl Rose, a screamed vocal like Pantera.
Maybe 1 is orchestral instruments
Maybe 2 is nature sounds
3 cartoon effects…
Something to that effect?
It would be very hard to encode a melody sequence of specific notes unless you had perfect pitch, and even still might be too cumbersome to easily implement.
I like the idea of sound playing a role in encoding though, it would just need to be worth it from a data encoding and speed perspective for me.
Interesting idea. If you yourself remember sounds well or even better than images it might be a good idea.
I personally account to the vast majority that remember per imaginary better. Empircally, although data is limited, this seems to be common, as hinted in my previous sentence. Looking forwards to updates.