KamHam <k.hamilton@thebackrooms.com>8:01 PM
to me ▼
Mike, I just got like a super genius idea. You know how our database contains a ton of videos with no time/date information, or just wrong info? And we get more videos like that each day. Well, I think I have an idea of how to recover it. Not sure if it actually works, but if it does it would be really cool. And it would only require for there to be the sound of fluorescent lights in the background.
In the Frontrooms, they have this technology called electrical network frequency analysis, where they use fluctuations in the frequency of a power grid they can hear in the background of a video (the hum of fluorescent lights is double the frequency of the AC) to timestamp videos by comparing the fluctuations to a database of such fluctuations. Now, we don't actually have such a database, at least yet, but I think we might be able to reconstruct some stuff from security cameras that have recorded audio. In the future, if we start building such a database, this will become more practical.
MichaelW <m.washington@thebackrooms.com>8:23 PM
to k.hamilton@thebackrooms.com ▼
I didn't fully understand your explanation to be honest. However, that sounds cool and useful. Though, there are some potential issues I see. First up, this idea assumes the hum of fluorescent lights would change in frequency, and that it would be consistent across different fluorescent lights. Seems unlikely, I can't see a reason for this. Also, for the idea regarding security cameras, it seems unlikely we'd have many that record audio, and I believe we regularly delete old security camera files anyways to make room for new ones.
However, I will look into this, because it has a lot of potential.
A Junior Operative attaching recording equipment to a fluorescent lamp in Level 4. He requested to remain anonymous.
After receiving authorization, we were finally able to begin work on the electrical network frequency analysis program. The goal of the initial stage is to record audio from several fluorescent lights in Level 4 and elsewhere. Using this data, we will be able to answer two questions posed by Michael W. at the start of the project: whether the frequency of the hum varies, and whether this variance is consistent across different lights. If the answer is affirmative, we can move to the next stages.
Setting up the recording equipment was surprisingly simple. We moved our retractable ladder around and sometimes climbed it to attach a microphone to a fluorescent light. All of our equipment is quite makeshift right now; the microphones are running on batteries and don't even transmit a signal, so we'll have to physically come pick them up later this week. It'll take quite a while to get the final results out.
This approach has certain limitations—for example, due to the low quality of the microphones, the data might not be fully accurate, or some areas might have other noises that obscure the humming. However, even then, it might be possible to see the pattern if it exists. And while I don't have evidence, I'd like to believe it does. I'd like to believe there's logic to this place and that it's something humans can figure out, even if both they and their tools are imperfect.
-Kamilla Hamilton, 03/11/2024.
KamHam <k.hamilton@thebackrooms.com>7:21 AM
to me ▼
This is big. We just analyzed the data, and it matches almost perfectly. Not just within Level 4—the recordings we got from microphones installed in other levels were also a match. The frequencies of the lights vary by time, and incredibly, they seem to vary the same way everywhere. Weirdly enough, it seems like we've finally found something here that's consistent.
MichaelW <m.washington@thebackrooms.com>9:19 AM
to k.hamilton@thebackrooms.com ▼
Wow, that's impressive and intriguing. Have you got any explanation for why it's so consistent? Seems rather odd since the data is from different levels and such. Also, have you tried this out on any old videos yet, like recordings or security cam footage with audio?
Regardless, I definitely think you've caught the attention of the people in charge with this project. It took a lot of work and time to get the initial trial funded, but things might get a lot smoother now.
KamHam <k.hamilton@thebackrooms.com>9:24 AM
to me ▼
No explanation yet, really. However, this would seem to imply the electricity that powers the lights is coming from some common source. Just speculation, of course.
Yeah, about the security cam footage, we tried to look for stuff with sounds, but couldn't find anything usable yet. As for short clips, we don't yet have the software to analyze videos, but we've got some plans and some expertise. Perhaps soon also support from the Tech Board.
Would really need it tbh. The flowers on my shelf are like WITHERING right now. No time to water 'em.
I'm happy to report we've finally been able to begin with stage 4 of the project. This time, we'll be connecting directly to the electrical network to measure its frequency, instead of the normal process of measuring the frequency of the audible hum of a fluorescent light and dividing it by two. Our new approach allows for greater accuracy time-wise and much less noise. We're expecting it to drastically improve the quality of our predictions.
In the image, you can see the first room being rigged up. The wire was plugged into our monitoring equipment from one end, and then carefully inserted into the electrical circuitry of a light from the other. The monitoring equipment then transmits a signal straight to Base Omega for convenient analysis. Anyways, the rigging was repeated for two more rooms for even more accuracy. We'll be comparing the output of the monitors with each other and our recordings and try to verify if everything makes sense.
Quick update: Using the new approach, we've already noticed that some of the individual pulses comprising the power seem to be larger in amplitude than the others. This is something for further study, as it could tell us about the origins of the electricity powering the lights.
-Kamilla Hamilton, 06/22/2024.
MichaelW <m.washington@thebackrooms.com>9:19 AM
to k.hamilton@thebackrooms.com ▼
Hi, how's the project going? Everything running as normal? Haven't checked up with y'all in a while.
By the way, did you manage to get the software developed? May I do a test?
KamHam <k.hamilton@thebackrooms.com>11:24 AM
to me ▼
Yeah, everything's running as normal (if you can call it that, still getting grey hairs from all the extra work). The readings we're getting from the raw electricity show even more correspondence than the humming does. The software is still in beta, but feel free to send me what you'd like to test. Not sure if we'll be able to pass the trial though.
MichaelW <m.washington@thebackrooms.com>7:49 PM
to k.hamilton@thebackrooms.com ▼
Alright, here we go. Let's say I was looking at the database randomly and I found this clip with a hum in the background. Do you think you can reconstruct when it was recorded from just the audio alone? I removed the metadata to make this more rigorous.
V406780.mp4
KamHam <k.hamilton@thebackrooms.com>11:53 PM
to me ▼
Alright, the analysis just finished. However, our confidence value is quite low. If the audio recording was longer, it would be higher.
Anyways, our guess is 07/14/2024, 5:21 PM, UTC. The unix timestamp is 1720966887. So, from about a month ago. Here are the graphs:
The frequency of the background hum in your video.
A match from our database from around a month ago.
Might go sleep soon so I won't get more of those grey hairs.
MichaelW <m.washington@thebackrooms.com>9:11 AM
to k.hamilton@thebackrooms.com ▼
Wow. Impressive. It's exactly correct. Good night—oh, it's already morning.
THE B.N.T.G. REVIEW
What we know, and what we trust.
Dozens dead in Level 11
KamHam <k.hamilton@thebackrooms.com>5:16 AM
to me ▼
Hello. If you've received this message, you are a higher-up involved in our electrical network frequency analysis program. We've made a very concerning discovery regarding the recent accident in Level 11. The following report is classified and should not be shown to people without proper clearance from the Tech Board.
MEG_classified_(3).pdf
In July, when we were analyzing the waveform of the electricity powering the lights, a pattern became apparent to us. One in about 28 pulses seemed to be larger in amplitude from the rest. However, as no potential explanation existed at the time, the fact was quickly forgotten after documentation. Below are two graphs highlighting the discovery.
Figure 1. The raw waveform of the electricity.
Figure 2. The major pulses algorithmically isolated.
A noticeable anomaly was present in the data picked up by our recording equipment on the day of the tragedy in Level 11. This anomaly was also captured as a lengthy flicker of the lights by several security cameras in different levels. By comparing our data and footage recently provided to the M.E.G. by the B.N.T.G., it is clear to us that this flicker coincided perfectly with the initial deaths resulting from the rapid structural collapse. When we inspected the waveform in more detail, we discovered that several major pulses happened almost instaneously. It was this, along with a large increase in the number of minor pulses, that caused the electrical wiring of the lamps to malfunction, unable to handle the sudden increase in frequency.
Figure 3. Graph of the major pulses at the beginning of the collapse.
A noteworthy detail to mention is that according to recent estimates, a human dies about every five ninths of a second, which matches closely with the usual rate of the major pulses. While we must avoid unnecessary speculation and treating unproven conjecture as fact, this apparent correlation between human deaths and the major pulses is rather concerning in light of our discoveries regarding the tragedy. I believe more research and funding is necessary to figure out the origin of the electricity powering the lights.
-Kamilla Hamilton, 11/27/2024.
