The Doorbell That Screams: I
Chapter One: The First Glitch
Accounts Vary is designed to be heard—it’s an experience. Let it speak to you, or enjoy the text below. (15 minutes)
Teresa, or Tess to anyone who still called, hadn’t planned to move back to her mother’s house. No one plans for that, do they? But after the stroke that left her mother weak and a little slow, the uncomfortable hospital stays, the unfinished renovations to live around, there wasn’t much of a choice. Her brother sent money from Vancouver, and her now-ex had gently suggested “the space would be good for us.” Tess stopped fighting against the chaos relentlessly hurled at her life and moved in.
This wasn’t where she grew up, but a home her mother, June, had bought 10 years earlier with the intention of DIY-ing in retirement. She’d accomplished a couple poor paint jobs, and some discount pallets turned into whatever Pinterest could imagine—shelves, a coffee table, outdoor seating so low it was a liability for anyone over 67.
The house was on a quiet street in the kind of suburb that may have been beautiful in 1998: cracked pavement seams, aging labradors on faded leashes, pale plastic deck chairs, and Christmas lights left through all four seasons. June’s house had a broken front step, a door that stuck, and a weird hum in the pipes that no plumber could explain. It was fine. Not haunted or cold, just tired.
To pass the time between freelance contracts and helping June, who stubbornly tried to do everything herself, Tess scrolled. There were all the usual suspects—social media, local and global news, blogs, streaming services, the saved recipes she’d never make. Three months after moving in, she joined Neighbourly, the local app that functioned as both a digital town square and a surveillance confessional. Most posts were predictable and dull: missing recycling bins, dog poop left on a lawn, suspicious vans that were probably food delivery, debates about lawn pesticides. But when the videos surfaced, well, she watched those on a loop.
Doorbell cameras, mostly. Grainy night footage of shadowy figures running, crouching, sneaking. People playing detective with timestamped stills, trying to enhance plate numbers or shoe brands. The kind of petty vigilance that made the world feel oddly manageable. Small crimes. Fixable problems. Many times, problems that weren’t really problems to anyone outside middle-class purgatory.
That’s how she saw the first one.
“WHAT is this screaming?? At 3:47 AM??? This is NOT okay. Ring caught it.”
Posted by user ShirlFromPalomino.
Attached was a short clip of just 23 seconds. Black and white video of a static porch, concrete steps, sad shrubs—no one’s front step, or everyone's. Then, the motion lights flickered on. A man appeared from the far left of frame, thin, mid-40s, standing still under the light. He didn’t knock. He didn’t move. His mouth open, silently stretched.
No sound.
The caption said “screaming,” but the video had been muted, supposedly for “privacy reasons.” The comments devolved quickly:
“Looks like drugs. Meth twitch.”
“Could be epilepsy? Alzheimers? Young but you never know.”
“Creep trying to mess with ppl’s cams. Weird trend for Tik Tok.”
Tess watched the video five more times. There was something held in the way he stood. Like the scream was being emitted, broadcast, not a typical rushing from the lungs in a burst of air and noise—it was suspended in his face.
She blinked herself back into the reality of the beige and blue guest room. It was hard to really make out the man's features, but he did look quite thin. Maybe he was sick, physically or mentally, or both. Maybe he belonged in the neighbourhood somehow, or maybe he didn’t. Yes, the timing was weird, and yes, it was unsettling. But it could be so many things.
Four minutes after locking her phone, Tess tried to comment, but when she pressed “send,” she got an error message. Then, the page refreshed, and the post was gone.
She went to the search bar and set the filter to users. ShirlFromPalomino no longer existed. Not a profile, a post, or even a greyed-out name associated with a comment. Nothing.
The next Thursday, it happened again.
Different post, different house. Same timestamp: 3:47 AM.
The man was closer this time—three feet from the door. He wore the same white windbreaker and the same deep furrows in his forehead. His mouth gaped open.
Tess took screenshots. The photo was still grainy, but he was closer this time, more distinct. Loading the image into a number of platforms, she reverse image searched the man’s face with no hits. Not on Google, not on Traceback. So, she cropped the windbreaker. There was a small reflective logo near the chest that matched an image on a now-defunct blog discussing vintage streetwear. The jacket was a long-discontinued model called the Comet, pulled from stores in the early 90s after reports of the fabric freezing stiff in cold weather. The blog was right, though; it was drenched in Y2K gym-class energy. Some 15-year-old somewhere would love it for the nostalgia they couldn't actually feel.
The searches took just a few minutes, and glancing at the clock, Tess knew she had time before June would wake for the day. Downloading the video to her desktop from the app, she right-clicked and clicked into “get info,” but there was nothing. It was stripped. Either the uploader had removed it, or the app did, which wouldn’t be uncommon. Many platforms remove metadata to protect users' privacy. She commented again. “Same guy as last week?”
But within an hour, before a response, the post was removed.
By the third Thursday, Tess had set an alarm for 3:40 a.m. She brewed tea quietly in the kitchen and wrapped herself in a quilt on the couch before opening the live map of doorbell pings on the Neighbourly desktop site.
3:47. Ping. New post, different house. Same man.
She screen-recorded it this time.
Two years earlier, Tess had done the production work for a small investigative podcast—one of those scrappy, true-crime-adjacent shows that relied on listener tips and voice memos recorded on bad mics. The show barely had the budget to pay her, so she used free tools to clean up distorted phone calls, sync interview tracks, and isolate background sounds that could end up being more revealing than the guests themselves.
She still had one of the tools—EchoView. It was clunky and slow, an old forensic audio app she’d downloaded during a late-night sprint. Not professional-grade or even all that user-friendly, but good enough to spot repeated loops or weird quirks hiding under the noise.
On first pass, it was blank. But when she boosted the gain and applied a high-pass filter to clear out sub-bass noise—the very low-frequency sounds that are felt more than heard—the silence began to crackle. The waveform wasn’t empty. It had shape.
She switched to spectrogram view—a heat map of the sound—and that’s where she saw it:
A smeared arc, glowing faintly orange, with sharp vertical spikes clustering around 130Hz. Human vocal range, but wrong somehow. Smoother than a scream, more sustained. No breaks. Just one long pull of sound across the middle of the spectrum, as if someone had exhaled through panic.
Beneath it, dimmer but steady, were repeating ripple patterns. Like an echo.
She exported the clip as a .wav file and uploaded it to Signal Archive, a public-facing anomaly database stitched together by ex-reporters and amateur audio engineers. It scraped FOIA releases, old emergency calls, and declassified military comms—useful for fiction podcasts, weirder for everything else.
One match.
A 1982 emergency dispatch recording from just outside Bellrock. Back then, it was a nearly-booming railway town near the edge of relevance—log freight, two churches, four bars. Now, nearly three hours from Tess, Bellrock was nothing but bent rail lines, overgrown access roads, and people who were too stubborn or too lazy to die anywhere else.
The report was labeled “No Contact Made.” A call came in at 3:46 a.m., made from a payphone outside an old water treatment station. The voice on the line was male, breathless. He claimed he could hear someone screaming from inside the pipes—said it had been going on for two nights, that he couldn’t sleep anymore. The scream was starting to sound like words.
The dispatcher kept him on the line for forty-seven seconds. Then the call cut out—no click, no goodbye. Just a slow stretch of static that warped upward in pitch, like a tape being chewed.
RCMP were dispatched an hour later. They found no one, but next to the runoff pipe near the tree line, was a cassette recorder. An old RadioShack model, still running.
The tape inside contained exactly 3 minutes and 47 seconds of screaming.
Not yelling. Not panic. Just a long, rhythmic sound like metal dragging across metal with, occasionally, the hint of a voice. A loop? A trick of acoustics? No one could say. The report noted “possibly animal in origin” and shelved it as a “miscellaneous disturbance.”
But Bellrock didn’t forget.
In fact, it was the perfect story for a town where not much happened and the most fun you could have was scaling the water tower or smoking bad, maybe oregano, weed in a forest.
The lore took on a life of its own as teenagers dared each other to go down by the pipe at night. Some said you could hear whispering if you stood quiet enough. One girl claimed her nose started bleeding when she got within ten feet of the drain. Another boy—Kyle something—went missing for three days and wouldn’t speak when they found him. Just wrote the number 47 over and over on his walls and bedsheets.
Over time, the legend twisted. Some said the man on the tape wasn’t screaming, he was laughing. Some said if you played it backward, you could hear a name. Some said the voice was never coming from the pipe—it was coming from the recording device itself.
Either way, as legends go, they gave it a name. They called him The Pipe Howler.
Every so often, the audio would resurface online—a Reddit post, a creepypasta, a weird chain email. But it always vanished just as quickly. The link would break. The file would become corrupt. The user account would be gone.
Tess sat at her desk, listening to the warped spectrogram she'd extracted from the doorbell footage—the same arc, the same frequency.
3 minutes. 47 seconds.
The scream didn’t just match the legend, it matched the timestamp on the Ring cam.
She sat back from her screen, her skin buzzing with something that wasn’t quite fear—not yet.
She opened the doorbell video again, and slowed it.
The man on the porch didn’t move. His mouth was open. The scream—if that’s what it was—should’ve looked strained. But his expression didn’t flicker. His face was perfectly still like a photograph stretched across time.
She checked the porch light. It flickered with the same rhythm as the loop. Six pulses, then a pause. Six again.
A pattern.
Tess compiled everything. She didn’t really know what she’d found—she wasn’t in the business of drawing hasty conclusions—she just knew it was odd. The people posting on Neighbourly, those watching them on a loop, may want to know. So, for the first time, Tess opened the app and hit “New Post.” It took her nearly 40 minutes, but she posted a detailed breakdown: timestamps, jackets, waveform comparisons, links to public records. She even included a mini FOIA guide for people who wanted to dig further.
By morning, the post was gone, and she had a notice that her account was being monitored for violating “Community Guidelines.”
That afternoon, she got a message on Signal.
No profile picture. No name. Just a single line: “It isn’t him screaming. It’s the doorbell.”
She opened the clip again. Paused it. Rewatched frame by frame.
This time, she saw it.
The timestamp froze. The “Ring” overlay in the corner twitched, stuttered, and for one frame, it simply read:
RUN.
Products, platforms and softwares mentioned in Accounts Vary will vary themselves. Like the details, some are real, some are fictionalized.