Between Bands

There’s a reason the radio station frequencies only end in odd numbers on FM and zero on AM. The technical reasons – the reasons they give on paper and publish “officially” – claim it’s to avoid interference from other radio stations and to give each broadcaster enough space to send all of their message.  It makes sense, for each person, company, or transmission to have their own dedicated “band” just for themselves. Everything in its place, and a place for everything.

Too bad the truth of it is much, much different.

In 1981, a team of researchers – comprised of amateur radio enthusiasts, HAM & CB radio operators, a few university fellows, and one senior technician from Bell Labs – began an initiative to comb through each and every band that they could to create a “noise profile” for the frequency spectrum. They thought that by isolating the background noise from each frequency, they could then remove it from their broadcasts to improve the strength and cohesion of their transmissions.

Their goal, in the end, was to generalize the process and patent it – essentially to prove it was possible and then make it available to use for everyone, for a small licensing fee.

At first, they didn’t see much success. Instruments at the time didn’t provide enough resolution or detail to read a single specific frequency without interference from nearby frequencies – the exact problem they were trying to solve, rearing its head from the get-go. Additionally, antennas are innately limited in range, so every time you moved the background noise would be different from before and the system would have to compensate for that.

They could filter out the frequencies using circuits, but sourcing the parts necessary for the entire spectrum would delay the team longer than they could afford. Then, inspiration struck – a European research team was rumored to be developing a digital radio system, which would allow conversion of analog radio signals into binary data that could be easily processed by computers, another new and powerful technology of the time.

The team heard about this and decided instead of focusing on encoding and decoding the frequencies to play well together, they would focus only on filtering them and processing the content left. It wasn’t revolutionary, but it was a step in the right direction. They now had a method to follow, and all that was left was implementing it and testing it.

With that, momentum began to build for the team. They had an early prototype within a couple of days, and were filtering and reading frequencies within hours of its completion. What they had built was essentially a wide band antenna attached to a computer, where the analog signals received would be digitally reduced to a single frequency, and the noise remaining on that frequency would be read and stored by the computer for other broadcasters to use.

There was no additional processing done on the analog data captured – it was just the raw signal for each frequency that their antenna could pick up. This meant that each broadcaster had to encode or decode that noise profile on their own, which of course meant that the research team had a new problem to solve.

With other teams now beginning their own efforts in digital radio broadcasting, they knew it wouldn’t be long before an official protocol was announced. So, rather than developing a sophisticated system from the ground up, they went the path of least resistance.

For each frequency, they took whatever was left over after filtering it through the computer and had the computer send the raw signal through a speaker. They thought that, at the very least, they would hear static or some kind of ringing, something that matched the frequency that they were isolating. Of course, the human ear can only process certain frequencies, and many of the frequencies they were testing were well outside that range.

They assumed for those cases that they would hear silence, crackling, or the kind of noise you would expect from tape recorders or vinyls at the time – perhaps artifacts from the computer processing that remained within the signal data.

For the audible frequency spectrum, their results matched their expectations perfectly. All that was left after filtering for each frequency point was a pitched ringing with intermittent pops or pauses, almost predictable in their regularity. Encouraged, they began to explore the frequencies beyond the audible band, the ones that they were really interested in.

The first frequency they tried – just a few notches above 20,000, the upper range of human hearing – they heard nothing but a faint buzzing, just loud enough to be the speaker membrane vibrating. They tried turning up the volume on the speaker as much as they were able to, and still nothing – just the faint hum of the speaker driver.

While this matched their expectations, they weren’t satisfied. There has to be something, they thought – it was only a few notches above ultrasonic, surely the signal wasn’t completely silent to human hearing. So, they looked to amplifying the signal digitally, as well as through the volume control on the speaker setup.

They went to the computer and after a few hours of debate and tuning, they had a program that would amplify the signal data without altering its content – making anything that was within the signal more “audible” to the output system they were using.  It was essentially a digital volume control that would work on top of their analog volume control, letting them boost the signal content even further than before.

By now, they realized they had created their own digital broadcasting system without intending to, but their focus remained on their original hypothesis. So, they loaded the signal content into the program that they created, amplified it as far as it could go digitally and manually, and pressed play.

For a second, there was nothing. Then the screaming began.

They shut it off immediately, thinking that something was wrong with their program. They reduced the amplification by half, assuming that starting at the maximum was their problem and they had injected too much external noise into the signal by trying to amplify it digitally. When they pressed play the second time, the volume was indeed reduced by half… but the screaming remained.

They couldn’t tell who or what was screaming. The voice – or voices - could have been male or female, young or old, but it was unending, a constant shriek that lanced through them like a cold dagger. They shut off the program, unnerved and bewildered, casting each other wary looks while glancing back at the frequency dial. 20,646 Hertz – just above human hearing.

A few wanted to shut down the project then and there – particularly the blue-collar researchers, those from non-professional and hobbyist radio backgrounds.  The remainder of the group elected to move forward, proposing that it had to be a glitch, that the other frequencies would show more normal results. So, they dialed it up a few notches – 20,650 Hertz, instead.

When they pressed play, it was as though a roaring fire had erupted within the lab. Several yelped, one researcher tripped over his chair, and two nearly fell over each other to turn the volume down. Even reduced to the lowest setting on both ends, the signal still came from the speaker as though an inferno was raging just meters away. A rushing, crackling whoosh, just missing the heat.

They let it run for several seconds before shutting it off, even more unnerved than before, but becoming more curious than afraid. They decided to test the resolution of their system, then – they would go to 20,651 Hertz, a difference of just one tick. Normal analog systems at such close range would experience some overlap, echoes of the neighboring frequencies. They hadn’t heard any traces of the first frequency in the second at a 5 Hertz difference, so perhaps their filtering worked, but it introduced its own noise that caused the… abnormalities in the signal.

There was no sound of blazing fire when they pressed play the third time. At 20,651 Hertz, the background “noise profile” that the team had went through so much effort to determine and isolate wasn’t background noise at all.

It was begging. Pleading; They heard a mix of men and women, young and old, crying out for help, for rescue. There was no hint or mention of from what, only save us, please, we need help, anyone, hundreds of unrecognizable voices speaking over one another in distress and despair.

They shut it off after only a few seconds, dutifully making notes even while they all turned pale and wide-eyed. There was no way their system had introduced those results as artificial noise – they had tested every component and profiled them exactly, entirely certain of what each one would do to the signal they received from the antenna.

Perhaps that was the issue, they thought – someone was already broadcasting on each of those frequencies, with startling clarity and signal strength, and they were picking up those broadcasts as their filtered background noise. The facility, however, was miles away from the closest tower, and they had sophisticated antenna arrays on the outside of the facility that let them know if there was anyone – or anything – transmitting in their general direction. Any broadcast at those frequencies they should have already known about, and their preliminary tests hadn’t found any at all.  

They paused the research while they recalibrated and examined all their instruments again. They told nobody – not at first – unsure if what they were finding was real or if it was just something wrong with their setup. After weeks of double and triple-checking each aspect of their system, however, they had no other answers, and no other choices.

They had to keep checking each frequency within their range, and they had to eventually publish their results for peer review.

When they turned on the system for the fourth time, nearly two months after their last test, there was only one person at the controls. The rest were standing or seated further back, watching with mixed anticipation and dread. A few were wearing ear protection – particularly the senior technician sitting at the control station.  They all held their breath as the technician adjusted the dials, setting the digital amplification at only 25%, and the frequency at 20,649 Hertz: one tick lower than the “fire frequency” as they had dubbed it, instead of one tick higher, which they had started calling the “disaster frequency.”

 The senior technician pressed play, and the noise that came from the speaker this time had everyone standing falling to their knees. The researchers gasped, unable to understand or comprehend the sound, but the unrelenting pressure of it, the cacophony that came from the speaker, even at a quarter amplification, drove them backwards.

The technician at the controls was affected the worst – but through ruptured eardrums and what turned out to be a detached retina he managed to turn off the broadcast and plunge the room into a blessed, ringing silence.

They, officially, had no idea what was going on by now – but they knew that they had to escalate this to the right people. Between the different members of the research group, they had several government and military contacts, and those channels would open up a few avenues that they previously couldn’t pursue until they had some sort of results. Given the sheer damage the fourth test had caused, they knew there would be interest.

So, they reached out and shared their preliminary results to a few select contacts across government and military agencies, pitching to them weaponization possibilities as well as counterintelligence and national security potential. The experience the Bell Labs technician brought to the table was crucial to this – as well as the testimony he gave after he finished recovering from the test’s aftermath.

Very quickly after this, the entire project became classified to the highest level, and publicly available information about the findings becomes unreliable.

The project continued for several years under the code name “Operation Whalesong,” with the cover story that it was exploring underwater communication technologies. Several of the junior researchers dropped out of the project or were quietly bought out of it, sworn to secrecy about their findings and placed in relocation programs.

The project quietly concluded without ever publishing its findings, but there are a few patents from that time, attributed to that research group and several government agencies, all centered around digital radio broadcasting; Specifically, the digital processing of ambient background signals and selective filtering of frequency-specific “abnormalities,” as well as the production and injection of artificial “white noise” into broadcasting systems to “reduce and remove innate abnormalities in frequency spectra.”

In simple terms, adding white noise and removing background noise, just like static on the television. And so, the team succeeded in its original goal – it found a noise profile for each frequency it investigated, and every frequency on the radio spectrum – but what they found was far beyond what they could have planned for.

There would be no licensing or selling their process. Instead, they ensured the technology would be given to everybody in the industry for free and integrated as quickly as possible, so that nobody could accidentally uncover the background “noise” that they had discovered. Fail safes were implemented, protocols were written, and the technology since then has grown by leaps and bounds without similar incidents.

But if you ever find yourself tuning into a radio frequency that’s just static, maybe just leave it at that and change the station. There’s a reason the frequencies are numbered and the volume only goes up so much, and it wouldn’t do to stumble across something dangerous – or worse – between the bands on the dial.