The full set of research logs I have kept regarding this investigation is now live on this site. Dating back weeks, these entries cover the files discovered via QR code scan, with location notes, timestamps, and any relevant audio observations included. Files are available for independent analysis via the Audio Archive.
At this point, all known codes have been documented. Every file has been processed and reviewed under multiple conditions. Phrase recognition appears to have plateaued. The most consistent vocal segments have been transcribed.
That said, several elements remain unresolved. Patterns do exist, but I am still unable to determine whether those patterns are intentional, coincidental, or the result of interpretive bias.
The possibility of matrixing continues to concern me. As mentioned in earlier entries, I’ve taken steps to limit expectation bias during review, but repeated solo analysis risks reinforcing misperception. A wider dataset of listeners may help clarify whether the patterns I have observed are consistent across perspectives.
To that end, the logs are now open for public review. If fresh perspectives are able to disprove my interpretations, I welcome that. If anyone sees something I have missed, I welcome that as well.
Moving forward, I will be posting further updates on this section of the website as they come.
At present, all files have been located, downloaded, transcribed, and analyzed to the extent currently possible using available tools. I have conducted multiple rounds of EQ testing, spectral layering, and cross-referencing across all audible segments. Phrase recognition has plateaued. No additional clarity has been gained in the past 72 hours.
All QR code placements have been mapped and georeferenced. No further sites have been identified.
The following remains unresolved:
- The origin of the files
- The identity (or consistency) of the speaker
- The conditions under which the audio was recorded
- The reason for thematic convergence
Despite the absence of direct continuity between files, certain patterns persist: tonal affect, cadence, language framing, and emotional posture. Phrases continue to center on loss, displacement, and agency. There is insufficient evidence to determine whether this is the result of editorial intent.
I have developed a provisional interpretation of the material that I am not yet prepared to document. It is based on contextual inference and associative logic rather than verifiable data.
Given the lack of new input, it may be appropriate to begin considering external analysis.
Public contribution could prove useful. Will reassess tomorrow.
I have spent most of today revisiting the previously documented vocal phrases, listening under both isolated and layered conditions, with fresh EQ curves applied. Particular attention was given to timing, cadence, and the plosive content of the consonants.
To review, audible phrases consisted of the following:
> “do you have the strength”
> “another victim in this…"
> “DO SOMETHING!”
> “a problem ignored, and then allowed to expand”
Each is brief, but the semantics remain coherent under multiple listenings. There is a commonality of tone: declarative, testimonial, even accusatory.
The most significant shift occurred while revisiting the Bridgewater file. The first to yield an intelligible phrase, and the one I initially logged as:
> “passed as a prism”
It has become clear to me now that this phrase was incorrectly transcribed. Clear as day, the speech cuts through as:
> “the past is a prison.”
What I once read as metaphor now feels more like statement.
With that revision, a thematic shape starts to emege. Consequently, I have begun to wonder if the phenomena captured in these recordings, and what they appear to focus on, may relate to their area of origin.
Today I reviewed field notes from a series of hikes I conducted between 2017 and 2019 in the area surrounding Hockomock Swamp. The primary focus at the time was cartographic: verifying discrepancies between local trail maps and satellite topography. No audio was recorded, and no EVP-related investigations were conducted. Still, in light of recent material, I thought it appropriate to revisit those notes and reconsider any overlooked relevance.
The terrain is inconsistent. Hard-packed trail gives way quickly to marsh and sedge. Water lines do not match any seasonal flood maps I’ve seen. GPS was unreliable past a certain depth, particularly in the northern interior. Consistent with many reports you'll find from other amateur researchers of various discipline and interest.
While this retread bore no conclusions, it allowed me to relive the sensation of being there. Subjective experience is normally something we try to account for and disregard, rather than necessarily lean into in these matters, but...
These sensations come to mind while reviewing the Bridgewater file. It is possible the location of the code had no relation to where it was recorded. But if terrain can affect signal, and signal behavior is influenced by environment, then a location like Hockomock - dense, humid, geologically layered - is at least worth factoring in.
It is well documented that this region has exhibited similar curiosities in the past. And there is, of course, the matter of historical context.
Tomorrow, I will return to the phrases found in prior files with fresh ears. I want to confirm my prior transcriptions.
Since completing the final scan, I have spent considerable time reassessing the material - not just the recordings themselves, but their placement. Reconstructing exact distances between stations is imprecise without GPS records, but the general sequence remains clear enough. The QR codes all appeared along the same commuter route, primarily on outbound platforms, most near posted schedules or seating. From a practical standpoint, this appears to have been a single route, marked in full. Convenience likely dictated the placements more than intention.
But intention may still be present elsewhere.
The first code I scanned (South Station) produced nothing of note at the time. A wash of static. I dismissed it, moved on. But the first recording to present anything intelligible, anything resembling speech, came from Bridgewater.
Bridgewater is a name familiar to anyone with even passing interest in regional folklore. Its associations with anomalous sightings, unexplained events, and electromagnetic irregularities have been documented, however unevenly, for decades. The Bridgewater Triangle has long been a magnet for speculation, though most accounts tend toward exaggeration. I have read enough of them to separate, in my opinion, rumor from true occurrence.
I have certainly spent my fair share of time in the area in the vain hope I would discover the undiscoverable, but I have never gone to the extent of bringing my own recording devices for such an end. Perhaps somebody else had, and perhaps their luck proved stronger than my own.
With the final QR code now logged, I have spent the past two days relistening to each file in sequence. I have continued rotating between playback environments, alternating between near-field monitors and open-back headphones, testing for perception drift. There are still faint details emerging... not new information, necessarily, but previously overlooked qualities. Tonal irregularities. Microphasing. Shifts in amplitude that don’t behave the same way twice.
What has changed most is not the content of the files, but the conditions under which I am listening. I know now what I did not then. I am no longer approaching these as isolated anomalies. I am hearing them together, in aggregate.
I have yet to reach a firm conclusion, but a few preliminary impressions are beginning to take shape:
Each file follows a similar structure, though none are identical.
Certain frequency bands recur across multiple files, particularly in the 1.7–2.2 kHz range.
Speech fragments vary in clarity and cadence, but most cluster within the first 90 seconds of playback.
Some endpoints taper off, some cut sharply, some fade beneath threshold.
These traits suggest coordination rather than coincidence. There is no metadata to confirm authorship or origin (not that Vocaroo could support such a thing) but the structural consistencies are difficult to ignore. I cannot yet determine if these were sourced from a single session, location, or signal, but I’ve stopped treating them as independent.
One working hypothesis is that these are excerpts from a larger whole. A single source signal segmented and distributed.
At present, I have made no meaningful progress in deciphering the more ambiguous vocal content. The phrases that cut through previously are as clear as initially described, if not more so.
This is the final recording to be formally logged, but the very first code I encountered. Now revisited with the benefit of experience and context.
Located at the far end of the outbound platform at South Station, the sticker remains affixed to the railing just before the last bench. This was the code that first alerted me to the presence of these recordings, and yet, at the time, I heard nothing. Having now reviewed eight other files, I’m left somewhat baffled by how much I failed to perceive during that initial encounter. Whether due to ambient interference, my own inattentiveness, or a simple lack of interpretive framework, I cannot say. But today, listening under controlled conditions and with a more attuned ear, the audio is unmistakably different.
The file is two minutes, sixteen seconds in length, hosted on Vocaroo like the others, and comprised mostly of static white noise, like the others.
Observations (Auditory):
00:19–00:23: “a problem ignored, and then allowed to expand”
02:05–02:16: abrupt reduction in volume and intensity of noise, fading into silence
The vocal line at 00:19 is clearly enunciated and cuts through the static without enhancement. It is calm, declarative. I am unable to determine if this voice matches any of the others present in the previous recordings. No processing was required to extract the phrase; it presented itself cleanly during the first focused listen.
Observations (Technical)
The waveform remains relatively even throughout the duration, with a gradual but distinct taper beginning just past the two-minute mark. There is no peak clipping around the phrase, and no compression artifacts suggestive of manipulation. Spectral data shows increased energy in the 1.8–2.1 kHz range during the vocal phrase, followed by a return to baseline. No unusual tonal elements were detected.
Now, hearing this in full, I find it difficult to believe I didn’t detect the phrase during my first exposure. It casts the entire project in a different light. Was this message always here, waiting to be heard?
With this entry, I have now documented and logged every QR code I encountered. Each file is archived, annotated, and prepared for further analysis. Whether this trail leads anywhere definitive remains to be seen.
The QR code at JFK/UMass was located directly to the left of the posted schedule on the outbound platform, above seating. Placement aligns with those observed at Campello and Brockton. The file accessed via the code is four minutes, eleven seconds in length, making it the longest recording in the series thus far. As with all previous entries, the file is hosted on Vocaroo and contains mostly static white noise.
Observations (Auditory):
00:16–00:21: choppy, repeating speech, consistent volume, unintelligible
00:50–00:54: choppy, repeating speech, consistent volume, unintelligible
01:07–01:11: choppy, repeating speech, rising and receding in volume, unintelligible
03:59–04:11: abrupt reduction in volume and intensity of noise, fading into silence
Despite the extended duration of the recording, very little additional material presents itself. All vocal anomalies are concentrated in the first 75 seconds, each displaying similar looping characteristics. There is no discernible message present in the audio, along the lines of something like the Bridgewater file.
Observations (Technical):
Spectral data confirms clustered harmonic activity in the 700–950 Hz range during each vocal segment, with moderate transient spikes above 2 kHz at onset points. Noise floor remains otherwise uniform throughout the file. Notably, there is no sharp termination - the file ends in a gradual descent, unlike the Quincy Center or Braintree entries.
While the extended length suggested a higher data yield, the recording has proven relatively sparse in new information. That said, the recurrence of choppy, repeated vocal fragments across multiple locations further supports the hypothesis of intentional structure.
With this entry, nearly all codes noted along my original travel route have now been documented. The only remaining code is South Station, the code that I initially encountered before beginning this more formal investigation.
Locating the code at Quincy Center required a more extensive search than anticipated. Nothing was visible immediately upon exit from the train, nor in the usual placement zones around benches or railings. After a thorough search of the inside, I found the sticker by the schedule posted by the bus benches outside the parkway. Aside from the Braintree code, this was easily the furthest I've had to venture, and hardest I've had to search, for a code thus far.
The QR code links to an audio file measuring two minutes, one second in length. It is hosted on Vocaroo and, like the others, lacks any metadata or identifying tags.
Observations (Auditory):
01:01–01:03: clear speech - "DO SOMETHING!”
02:00–02:01: abrupt cessation of all static and other sound, file ends
The delivery on the audible speech is forceful and unmistakably human. The phrase cuts through the static with full clarity — no filtering or enhancement required. There is brief peak clipping in the waveform, indicating either an uncompressed capture or an intentional spike.
Notably, this marks only the second instance of the static being cut off in this manner, and the only time this coincided with the end of the file.
Observations (Technical):
Spectral analysis reveals a sharp harmonic spike accompanying the vocal burst at 01:01, centered around 2.8–3.2 kHz, with faint overtones reaching above 6 kHz. A brief suppression of broadband noise occurs just before the outburst — roughly 300 ms of reduced energy between 00:59.7 and 01:00.0 — though it’s unclear whether this is mechanical gating or deliberate shaping.
No 'musical' elements are present, unlike in the Braintree file. The waveform remains relatively uniform until the final second, where an immediate amplitude drop confirms the hard cutoff.
This recording will be revisited after transcription refinement. The rhetorical demand in the vocal line feels targeted, but the question of matrixing once again rears its head. Whether that perception is psychological or programmatic remains to be determined.
Tomorrow: JFK/UMass.
Unlike prior locations, I had not spotted a QR code here during the commute on which I spotted the others. However, given the consistent placement of codes at all prior stops along this line starting at Bridgewater, I proceeded under the assumption that one was likely here. After a few minutes inspecting the departing platform, I headed down to the parking garage. Before reaching the garage, I found a QR code affixed in a position consistent with prior placements: beneath the large BRAINTREE signage above a bench, right along the leftmost section of the white portion.
The file accessed via the QR code is two minutes, forty-two seconds in length. As with all prior files, the majority of audio contained was white noise.
Observations (Auditory):
00:21–00:24: faint echoing sound, possibly mechanical in nature
01:02–01:09: *“why will you not wake me up”?*
02:06–02:36: faint sounds reminiscent of a piano
02:38–02:42: abrupt cessation of all static and other sound until end of file
The phrase detected at 01:02 is the clearest linguistic articulation observed in any file (on first pass) thus far. The sentence is fully structured, with both subject and interrogative verb intact. I listened via both near-field monitors and open-back headphones to account for differences in format. Formant spacing, plosive articulation, and stress cadence are all suggestive of natural speech. This will be tentatively marked as a Class B EVP, pending phonetic mapping.
Spectral analysis confirms irregular harmonic clustering in the 200–600 Hz range during the piano-like passage between 02:06 and 02:36, with faint spikes occurring at consistent intervals. The musical tonality is subtle, but persists long enough to be considered deliberate.
Of note: this is the first file where static fully and abruptly ends before the file does. From 02:38 onward, the waveform is flat. This differs from previous files where the static fades into eventual silence.
Given the strength of the spoken phrase, I will revisit the earlier recordings using updated EQ and noise-reduction profiles when time permits. If this message is part of a larger sequence, even more questions are raised.
Today’s documentation was conducted at Holbrook/Randolph Station. The QR code's location was almost identical to the preceding one: to the left of the posted schedule on an outbound platform. Whether this has any relevance to the files themselves is not readily apparent. The linked file is two minutes, fifty-one seconds in length. As with previous entries, the recording appears to consist entirely of static white noise. However, this is the first file logged after my decision to implement spectral visualization into my workflow.
Observations (Auditory)
00:11-00:24: choppy, fragmented, repeating speech, rising and falling volume, unintelligible
00:36-00:41: choppy speech, sustained volume, unintelligible
01:04-01:20: choppy, fragmented, repeating speech, rising and falling volume, unintelligible
01:21-01:31: uninterrupted speech, rising and falling volume, unintelligible
01:59-02:02: "do you have the strength"(?)
02:41-02:51: abrupt reduction in static volume, fading into silence
Observations (Spectral):
Using Spek, I processed the full waveform and isolated the above segments visually. The file contains pronounced density clusters in the 1.2–1.8 kHz band — especially between 01:00 and 01:30 — which coincide with the longest uninterrupted segment of speech-like noise.
The phrase detected at 01:59 (“do you have the strength?”) is tentative but notable. The fricative trails and sibilant structure are consistent with speech patterns, though the formants are warped. I have marked this as a possible Class B EVP, pending confirmation from spectral overlap in other files. No prior recordings beyond the Bridgewater file (which I have also upgraded to Class B upon further listening), have shown such direct semantic clarity, so caution remains warranted.
Tomorrow’s target is Braintree Station. I will continue layering auditory analysis with visual mapping.
Today’s documentation was conducted at Montello Station. The QR code was located directly to the left of the posted schedule on the outbound platform. The file is two minutes, fifty-nine seconds in length. As with all prior recordings, it consists entirely of white noise.
Observations:
00:25-00:37: faintly audible speech barely cutting through static, sustained volume, unintelligible
00:53-00:57: faintly audible speech, rising and lowering in volume, unintelligible
01:04-01:07: faintly audible speech, sustained volume, repeating word? unintelligible
01:38-01:43: faintly audible speech, sustained volume, unintelligible speech
02:13-02:19: some noise that sounds mechanical? repeated, higher-pitched, almost glitchy
02:48-02:59 abrupt reduction in static volume, fading into silence
At this stage, I have refrained from applying any audio processing or enhancement. Observations remain based solely on critical listening and timestamp logging. No coherent phrases have yet emerged beyond what was detailed in the Bridgewater file. The mechanical element at 02:13 is of particular interest, though I will withhold conclusions until more data is gathered.
The repeated inability to extract meaningful information from these files is becoming increasingly frustrating. It is now my assessment that further progress via critical listening alone is unlikely. In returning to the Bridgewater file this evening, I heard the phrase “passed as a prism” more clearly than before. There is a certainty to it now. My interpretation has not changed, but my confidence in it has. The phrasing, rhythm, and consonant shape remain consistent across multiple listenings. This suggests a pattern, not matrixing.
In light of this, I will now begin applying more advanced forms of analysis. Starting with spectral visualization tools such as Spek, and progressing toward more intense methodology if necessary. Though I remain cautious, the possibility that these files are part of a coherent, encoded transmission can no longer be dismissed as fringe speculation.
Today’s documentation was conducted at Brockton Station. The QR code was located to the left of the schedule posted along seating on the outbound platform. The file is two minutes, three seconds in length. As with all prior recordings, the audio appears to consist entirely of white noise.
Observations:
00:10–00:13: inaudible speech, sustained volume, unintelligible
00:19-00:25: inaudible speech, sustained volume, unintelligible
00:41-00:48: inaudible speech, sustained volume, unintelligible
01:32-01:35: inaudible speech, sustained volume, unintelligible
01:53-02:04: abrupt reduction in volume and intensity of noise, fading into silence.
At this stage, I have refrained from applying any audio processing or enhancement. Observations remain based solely on critical listening and timestamp logging. No coherent phrases have yet emerged beyond what was detailed in the previous files.
Next station on the line is Montello. Visit planned for tomorrow.
Today marks the first of what I anticipate will be several field documentation efforts. My aim is to revisit each station where I recall seeing a QR code and formally log each recording in sequence. I began at Campello Station in Brockton, MA. The sticker was still affixed to the right of the parking section arrow above seating along the outbound platform.
The QR code links to another Vocaroo-hosted recording, this one lasting two minutes, seventeen seconds. Like the Bridgewater and South Station recordings before it, this file appears to consist entirely of white noise.
Observations:
00:16-00:18: fragmented speech of sustained volume, unintelligible (almost sounds like "an evil", but inconclusive)
00:28-00:38: more fragmented speech, sustained volume, unintelligible
01:09-01:16: fragmented speech, rising and falling volume, largely unintelligible (almost sounds like "half-life" or (have life", inconclusive)
01:33-01:37: "another victim in this (???)"
02:06-02:17: abrupt reduction in volume and intensity of noise, fading into silence.
I have flagged this entry for future comparison once more of the files are logged. If I can establish a recurring motif in these upper registers or in volume shaping, it may justify the use of batch spectral analysis.
Tomorrow I will proceed to Campello’s neighbor: Brockton Station. If these stickers were placed in sequence, the next code should be found there.
Further examination of the Bridgewater file has yielded a partial breakthrough. At approximately 01:11 into the recording, I identified what appears to be a coherent phrase:
"passed as a prism"
Though distorted and partially obscured by static, the vocal cadence, consonantal structure, and rhythm of the segment support the interpretation. No enhancements were applied prior to detection. The phrase emerged during standard playback under quiet listening conditions using Sennheiser HD 600 headphones.
I returned to the full recording and logged additional timestamps of possible interest, but no phrases of equal clarity were identified during this session. The remainder of the file continues to fluctuate in amplitude and noise density, particularly in the final 30 seconds.
As mentioned in prior entries, I consider the Bridgewater file to be distinct from the South Station recording in both behavior and structure. Whether this is due to differing intent or differing source conditions remains undetermined. However, in light of the above findings, I believe it is now reasonable to proceed under the assumption that these recordings may constitute a series. Considering the ever-present risk of matrixing, I will be treating this as a tentative Class C EVP.
Beginning tomorrow, I will return to the station where I first noticed one of these QR codes and begin formal documentation and scanning of each individual sticker found. This will allow for clear sequencing and prevent auditory fatigue from biasing my assessments.
If this is a coordinated project or message, establishing chronology and content range will be essential.
Examination of the Bridgewater file has yet to produce concrete results, though progress is being made. The recording remains, in its base state, a mass of indistinct noise. Following my consideration yesterday that this could be a suspected EVP, I began treating the audio the way the circumstances would dictate. In the calm of my study, I relaxed myself. Well-rested, with sufficient time spent away from any number of potentially contaminating influences or interactions, I revisited the file.
At 00:28, I detected an anomaly suggestive of human vocalization. The shift was a brief but unmistakable fluctuation in the noise floor, followed by what I can only describe as the shape of a voice. No words can yet be extracted, but the cadence and compression patterns were noticeably different than the surrounding static.
Between 00:46 and 00:52, I marked a cluster of what sounds like looped or stuttered speech. Using Audacity, I isolated this segment and experimented with tempo-stretching and bandpass filtering around 500–1500 Hz. The result is still unintelligible, but the contours resemble fragmented syllables.
From 01:02 to 01:19, a similar pattern, choppy bursts of sound that rise and fall in volume, emerges. Here too, I attempted dynamic range compression and reverse playback, with no breakthrough. Still, the irregularity stands out from the baseline static.
At 02:37, the static begins to reduce in volume. This reduction continues until the 02:59, at which point all sound ceases. No vocal presence accompanies this change, but the tapering comes off as intentional. Whether this was by design from the recording parties or not, I am unsure.
These timestamps mark the strongest anomalies so far. I've looped each at length, both through semi-open headphones and studio monitors, with no new patterns emerging. For now, I have refrained from spectral analysis, wanting to first engage with the recording in its “native” form.
Beyond the obvious differences in file length and URL, I am now convinced that this Bridgewater recording is a separate beast entirely from the one linked to South Station. Having said that, this further inquiry into the Bridgewater file does pose more questions. Have I overlooked something in the South Station file? And what of the other codes I had seen along my commute? I distinctly recall one at the JFK/UMass stop, and logic dictates there could be more I overlooked. Come Monday, I may have to keep my eyes peeled even more than usual.
This journal will function as a record of my investigation into audio recordings sourced from QR codes posted throughout train stations in Southeastern Massachusetts. For the moment, this documentation is purely for my own personal use. However, should this investigation result in a discovery of substantial value, this record could be of use to the scientific community at large. With that in mind, I will aim to keep this record as detailed and objective as possible.
Yesterday, I noticed very small stickers of a QR code posted across different stops along my regular weekday commute. At my destination, spotting yet another sticker like the others, I scanned it. This code was found at South Station in Boston, Massachusetts. The code linked to an anonymous file upload site called Vocaroo, hosting an audio file lasting two minutes, twelve seconds. The audio consisted of white noise, nothing more.
Today, finding myself in front of another of these stickers, and with my curiosity getting the better of me, I scanned the code again. As it turns out, this was a different code, linking to a different audio file - this one lasting three minutes, sixteen seconds. Once again, the audio consisted of white noise, with no discernible message or content. However, I noticed traits in this file that were not present in the other. Noticeable fluctuations in volume and intensity, as if the noise was a result of signal interference. Returning home at the end of the day, I listened to the audio a few more times in better-isolated conditions and have confirmed that this file is behaving in a way the first did not. It was then that I considered the possibility that this was a field recording, an attempt to capture white noise EVPs. Why this recording would be shared in such a cryptic manner, by uncredited parties, devoid of any context and without any clear method for response from the public, is not immediately clear. Further analysis is warranted.