Holes in ad's spectrogram


Hey I don’t know if it has been said elsewhere but I cannot seem to find it. All of the ad since the Waking Titan first broadcast message contain little squared hole in the spectrum (image will follow). They are the same length and on the same “pitch” in one message, but between them all different.

(All value are approximation as I had to draw to see number)
The first broadcast:
Hole length: 214 ms
Pitch range: 454.75 Hz to 629.76 Hz (175.01)

Echo Ad:
Hole length: 214 ms
Pitch range: 485.06 Hz to 686.39 Hz (201.33)

Multiverse Ad:
Hole length: 214 ms
Pitch range: 1244.9 Hz to 1449.5 Hz (204.64)

Myriad Ad:
Hole length: 214 ms
Pitch range: 627.43 Hz to 781.84 Hz (154.41)

Superlumina Ad:
Hole length: 214 ms
Pitch range: 1194 Hz to 1728.2 Hz (534.17)

Again don’t be picky on those number, little drawing error could move between ±20 Hz easily. But the superlumina range is realy bigger than the echo one. The audio file is those given in the gamedetectives.

So yeah, I thing there is something here, but can’t figure out how make sense of this, maybe a kind of code, but is it one letter by ad ? Or if it’s more than one letter how to encode the message within one variable (time base).


Assuming my math is right. Sending 1500 bytes as a stream of data across a 56k link would take 214ms.
This is very close to the Ethernet frame size of 1518, if there was an Ethernet frame encoded somewhere…

bytes to bits
(1500*8)/56 = 214ms

(1518*8)/56 = 216ms


I thought it was odd when I spotted those black squares/punch holes at the lower sections of the spectrogram in the latest ad but to see that it’s been happening in all of the ads leads me to believe it’s intentional and not just because of a mishap with sending audio over the net (all the music and other adverts had no “squares” at the bottom during multiple recordings I was trying to get).

On top of that, the latest ad had an odd sine-wave in its spectrogram. I would considering it out of place but it’s too specifically tuned.


I had been curious about the ads in their entirety . Is that morse code again?
I feel like the ads could even be placed together in some way to get something. Whatever that may be. Although to have something complete, perhaps we need to hear the rest of the ads. And the project-wt site has quite a few stations still active.
Could be something to wrap up the stations when they’ve all been used?


Or binary. Where the lines close to one another are 0 and the singled out dots are 1? Idk, just spitballing…


@Espilonarge suggested on Discord that these could be holes in a computer punch card. If you overlaid all the videos and isolated the holes in each, it might look like one.


Maybe I’m crazy, but I’m seeing distorted faces in Shark’s spectrogram and it’s really creeping me out. Myriad and Echo are especially bad.


“Thanks for taking participation on our newest Rorschach Test. Give me a short moment to evaluate the results… Ok… Hm Hmmm… Seeing Ghostface, no… Seeing Birds, no… Seeing the Devil, no… Ahh, here we go, seeing distorted faces: You are afraid of slowly becoming a Cookie and being eaten by huge pink robot bunnies while listening to Daft Punk - Technologic. Don’t worry! That is absolutely normal while taking part in WT”


If you look at the https://www.multiverse-75.com/ website the banner has one of those squares in it also. Wonder if you line it up or overlay the spectograms on them you get anything?


I’ve been perplexed by this myself. Cassette 9 of 16 4th Glyph wiki uses frequencies to convert to MIDI to finally convert to numbers and letters to get the answer WBHEARL. This tells me it’s not beyond them to hide other code into these audio ads. Looking back at the math required to get that glyph password, it appears they got it wrong, the MIDI numbers should have been 72,50,56,53,49,67,60 which then neatly equates to 37,15,21,18,14,32,25 and then 10,15,21,18,14,5,25 and equals JOURNEY. I think the password was bruteforced first, and then they backwards engineered their math to suit it. Very suspicious.

But anyway, each black square in the ads is on a different pitch range, they are also repeated in irregular fashion. This means it’s more complicated than just applying a scale. The spacing of the squares seem indicative of a secondary encryption, but is too basic for morse.

If we take the pitch ranges you’ve noted, they roughly equate to:
-1st clue ad MIDI 72
-2nd echo MIDI 74
-3rd multiverse MIDI 89
-4th myriad MIDI 77
-5th superlumina MIDI 90
Using the same chart from the 4th glyph eventually gives me seemingly useless letters, unless someone else wants to try please. We might need more precision with the pitch range numbers to know for sure.


Yeah, the little innocuous white square on Multiverse has bothered me to. It doesnt do anything, but it’s there .


The holes remind me of these:


I’m pretty sure it’s left over from the WordPress theme. If there were multiple images in the header background thing, that would be where you could cycle through them.


Okay, just for the sake of ruling out possibilities without blindly ignoring them i made a compilation image of the spectrographs. Important Notes: The frequencies of all graphs are adjusted to fit each other. The length in seconds is adjusted for all samples by stretching the timescale. The squares itself are just a rough cut-out, which means the size of the squares is in no way meaningful in this image except the differentiation into long sqares and short squares.


The most dubious for me was finding two squares in the upper row which overlap each other both before and after stretching the timescale. However it is interesting that all squares fit in a timeframe from approximately 0:00 mins to 0:27 mins even though some ads were longer.

My conclusion for now: Unless there is some system that translates data of this kind into something meaningful, or there will be more ads which provide new light into it, or it is significant that someone tried to build a city in a audio file, overlapping is off the track…

Edit: Sorry for the bad visibility, but i am in a rush and a white background layer was the fastest solution.


I agree the solution was based on back engineering the already found solution, which had been changed after. You can see the bit of discussion I had about that here:

In all honesty, I found my solution to make more sense. In the end it did not matter much, but I prefer the actual logic behind a solution over an engineered solution based on a known outcome.


Looking at the squares, that reminds me quite a bit of MIDI sequencer data. Unless these squares are broadcasting artifacts, they seem deliberate.


I overlayed the squares of all 5 ads so far in a basic tone scale:

It Almost sounds like something melodic, even without an Atlas-65 ad included.
🤷 ¯\(ツ)