Audio Engineering: Paranormal EVP Analysis – “Don’t hurt me”

Paranormal EVP Analysis: “Don’t hurt me”

Background

We were contacted by YN, who submits this suspected EVP for analysis and evaluation.

 

This is what the YN believes is on the recording:

Don’t hurt me!
You’re a slave.
You’re a slave.
Don’t hurt me!
You can have whatever you want.
Please stop—

YN states further states:

“On October 21st, 2008, after leaving a night-time college course for high schoolers, my friend (NAME WITHHELD) and I decided to head to Washington Park on 1st street at the border of Union City and Jersey City, NJ. This would be a 18-block walk as we were coming from 18th street. While crossing 3rd St on Bergenline Avenue, we thought it would be fun to record a “parody” of Quarantine which was a (really bad) horror movie we had seen that past weekend.”

“That is why I ask for the time (approximately 9:08 PM); because there was a similar scene in the movie. Later that week, (NAME WITHHELD) and I decided to watch our parody video. It was then that we noticed the voice of that girl. We were completely shocked!! When we were recording, the street was pretty empty. There was no one on our side. No kids playing, no one passing by, nothing. Just me and (NAME WITHHELD) which is why we stopped to do the recording or else I would be too embarrassed to do something so silly in the public eye. Even so, something like a girl screaming for help would have certainly caught our attention.”

“Furthermore, we couldn’t believe how clear it was and the different tones to it because every EVP we tried looking up was always very hard to understand and filled with static. We tried doing some online research that would give some history to the EVP but we didn’t find much. We tried going back with the camera (and a couple of other enthusiastic friends) but didn’t pick up anything new.”

“[This is] a 16 second video trimmed from a 3 minute video . . . . The video is exceptional in the sense of how long the EVP is as well as how clear and well defined the voice is. I’ve never heard anything like it before anywhere.”

“This was recorded on my friend’s Sony NSC-GC1.”

PRELIMINARY EXPLANATIONS OR ASSUMPTIONS

YN states:

“After I heard the EVP, I tried doing some online research but I couldn’t find anything specific. The only thing I know for sure is that that area of New Jersey (Bergen County and Hudson County) had the most slaves in the state. New Jersey also didn’t abolish slavery until 1804, 21 years after the American Revolution, with slaves still existing in the state until 1864. Four underground railroad routes ended in Jersey City as well. So I think it’s definitely possible that at some point there may have been an altercation between a slave and a young female in that area. Maybe one escaping?

“Most people conclude that it was the group of people across the street. I tell them it’s impossible because it would have caught my attention like in other parts of the video (ie, 2:12 and 3:03) and the voice would have sounded too distant.

“Another explanation I’ve heard was that there was a child playing nearby. I usually then point out that I am facing the camera, therefore I have a view of what is behind my friend, and my friend has the camera on me which shows what’s behind me, and I assure them that there was no one on our side of the street from my viewpoint or the cameras. I also think that the “ambience” of the sound doesn’t match the background noise that you would normally hear from someone on the street (like the man talking on the phone at 1:15 and the people playing basketball at 2:12).

OUTSTANDING QUESTIONS

YN states:

“What would be great is if the analysis can show that the voice couldn’t have come from across the street or behind a person because of certain qualities of the voice.”

“Some have a hard time hearing “you’re a slave” etc and think it is saying something else. Is there a way to analyse the exact words based on the frequencies, etc?”

“At about 12 seconds into the video, I believe the voice says “Please stop–” and from there it becomes too difficult to understand. Is there any way to clarify what she’s saying? Also, does it truly end there? Or does it actually continue at a lower tone as we begin to walk?”

“At about 2-3 seconds into the video, there is a slight high pitch sound that I also didn’t notice when recording. To me it sounds like a cat meowing and it occurs right before the voice starts speaking. Is there a way to check the frequencies to see if it is consistent with a meow? If so, that would be even more intriguing considering I’m a huge animal lover who never misses a chance to approach a street cat so I’m wondering if that is also part of the voice.”


Our Analysis

Sonic Quality

The suspected EVP’s sonic qualities are exceptionally strong and clear.  The voice is consistent with that of a female/child English speaker. The voice frequencies most detectable range from ~600Hz to ~8-10kHz, where vowels and consonants are most intelligible. However, the lowest detectable frequency is ~250Hz, consistent with the average fundamental frequency for a female voice (~200Hz) and a child’s voice (~300Hz). Harmonics are detectable as high as 18kHz.

The suspected EVP’s audio level is lower than than that of YN and his friend, and it has a greater reverberant quality to it–consistent to a voice being recorded in proximity to but not in the near field pick up zone of the microphone, perhaps 25 meters away or more.

The camcorder (Sony NSC-GC1) operator has no control over the audio input, which is a built in mic on top of the unit. The sound file provided is dual-mono, without stereo-directionality.

Wave Form

The suspected EVP appears in the audio wave file to be captured (recorded) consistent with technical parameters of the native microphone.  In other words, the suspected EVP appears in the wave form to be recorded consistent with the voices of YN and friend. It does not appear to be influenced by electro-magnetic interference (EMI), internal electronic noise of the camera, nor does it appear appear to be a modulation of an extraneous carrier wave.

Furthermore, the suspected EVP does not appear to be added to, dubbed, spliced, or otherwise inserted into the audio file post recording. The wave form appears to have no obvious DC offset or “zero-line” voltage markers that might signal dubbing.

Intelligibility

Voice isolation – Completely isolating a voice within a mix of other voices is not an easy task–if not impossible or desirable–because altering audio dynamics always affects a voice’s unique make up of fundamental and harmonic frequencies, all of which combine to give it its distinct sonic identification (i.e., timbre).  The voice in this clip could not be isolated, as say voices recorded on sperate tracks can be isolated.

However, we were able to manipulate the audio to somewhat improving intelligibility, a technique that sometimes helps a voice stand out against other voices or background sound.

Here we manipulated the audio by cutting the timeline to include just the essential audio.  We applied spectral noise reduction to remove as much ambient background noise as we could without affecting the voice.  We applied both high pass and low pass frequency filters to remove unnecessary frequencies at the top and bottom ends of the spectrum.  We minimized the voices of YN and friend by reducing their fundamental and identifiable harmonic frequencies by 10dB. We reduced the voice’s initial start frequency by 25% (from 512Hz to 384Hz) to give it more bass presence. Finally, we reduced the track’s tempo by 10%.

Here is the original clip in isolation:

 

Here is the manipulated clip:

 

We are not certain that these manipulations help us better identify what is being uttered by the voice, but they do reinforce our initial thoughts about intelligibility.  For sections of the suspected EVP, the audio quality is exceptional and unusually clear. The voice is intelligible for the most part, except where YN and friend talk over the voice.

We asked members of a paranormal investigation group to listen to the recording and interpret for themselves what the voice might be saying. We did not give the group the transcript provided by YN, but we did mention the location and possible connection to slave times. Here are some responses directly from the investigators:

RK: Girl cries: “Don’t hurt me don’t hurt me nurse something don’t hurt me.”

MRH: Wow. It’s hard to understand with the talking but definitely an agonizing cry , then don’t hurt me … then one or two words I can’t make and then please don’t hurt me. Don’t hurt me. I agree with RK that it sounds like nurse … This is what I hear…don’t hurt me, nurse green, nurse green don’t hurt me, give me my doll it’s mine. Please don’t hurt me.

CH: Can’t really say its a slave per se but it is a child, female from the sounds of it quite clearly, sounds like “Don’t hurt me, don’t hurt me you’re a slave, you’re a slave, you can touch me all you want but please don’t hurt me.” Honestly it does not sound like it’s coming from a “slave” but does sound like perhaps a master’s child if it’s actually the case there. Either way it is a clear female child stating “don’t hurt me.”

I wish we had more of the clip … it’s quite rare to get a voice that is that very clear but I have had it happen so I am going to assume that this is legitimate until I have reason to disbelieve… 

KB: Sounds TOO clear to me. Questioning it’s authenticity.

HE: I think I might agree [as to its false authenticity] . But I’m just not sure. It would, for me , depend on the source.

HR: Don’t hurt me, I hear.

SMA: I hear don’t hurt me nurse please nurse please give me back and then I don’t know what she says after that but then she says please don’t hurt me and it definitely sounds like a little girl.

Evaluation

The voice of the suspected EVP is conclusively that of a female or child. For the most part, segments are clear and highly intelligible, more so than most EVPs.  However, there are sections that, to our ears are unintelligible and we can offer no determination as to what the voice is saying where there is voice overlap with YN and friend.  We clearly understand, “Don’t hurt me” and “please don’t hurt me.”

As for meaning, we can only be certain of what we can clearly and reasonably understand. Whenever there is intelligibility, there is the risk of pareidolia–the tendency to perceive something as meaningful based on nebulous auditory or visual stimuli (e.g., faces appearing in the clouds) and/or willful outcomes (e.g., that face in the cloud looks like someone’s face we’ve been hoping to find in the clouds). We always err on the side of clear and reasonable intelligibility.

Linguistically, “Don’t hurt me” is an imperative, second-person statement. The speech act is exclamatory and pleading, suggesting there is an urgency or crisis taking place. However, there is no linguistic marker that helps us understand if the utterance is directed to an individual or group.

As for the utterance directed to “a slave,” that interpretation depends on whether one hears the voice as saying, “you’re a slave.”  For us, both intelligibility and meaning are indeterminant. However, it is interesting to note that the phrase is repeated, linguistically marking it as something the speaker clearly wants to underscore. In rhetoric and discourse studies, this is called epizeuxis–the repetition of a word or phrase in immediate succession, typically within the same sentence, for vehemence or emphasis, as well as emotional appeal.

 But to assume for a moment that the voice is declaring, “You’re a slave,” we might reasonably infer, in the absence of any other evidence, that the woman or child speaker’s statement is to differentiate herself from the person to whom she is speaking. Does this mean that the female is not a slave? A white person? Perhaps even a free black female? We just don’t know for certain.

As for the phrase, “you can have anything you want,” this too is unclear to us and we can offer no interpretation.

As for the appearance of the suspected EVP itself, it is reported to be unsolicited.  Such phenomena are not unheard of, but they seem to be uncommon according to paranormal research. Furthermore, the suspected EVP appears to be dissociated from YN and his friend, their presence on the street, or what they were doing there. In other words, it appears that there is no connection whatsoever between the suspected EVP and the recordists.

Conclusions

The voice heard on the recording is real.  However, it is not an EVP, but an AVP (acoustic voice phenomenon).

The voice is intelligible and meaningful.  As pointed out, there are remarkably intelligible segments of the voice, but there are also unintelligible segments as well.  “Don’t hurt me” and “please don’t hurt me” stand out as the most clearly discernable phrases. The voice is clearly emotional, frantic, and directed toward someone or multiple persons. The unintelligible phrases, naturally, are of no help in understanding the complete situation or context that evokes the voice’s desperate plea.

The origin of the voice is inconclusive. From an audio perspective, the voice recording exhibits all the technical markers of being recorded on the street via the camcorder’s microphone in proximity to YN and his friend.  However, its origin cannot be explained in light of YN assertion that the voice was not present on the street while he and his friend were recording their video. In this regard, our technical findings relative to the origin of the voice are at odds with YN’s assertion. We let our technical findings stand on their own merits, but we do not doubt YN recollection.

This is perhaps the central mystery of the suspected AVP.

In response to the Outstanding Questions listed above:

“What would be great is if the analysis can show that the voice couldn’t have come from across the street or behind a person because of certain qualities of the voice.”

Response: Although not verified it is common for camcorders such as the Sony NSC-GC1 to use powered or electret type (permanently charged) condenser microphones, which are very sensitive and capable of picking up very low audio signals, such as voices at considerable distances from the microphone. As such, it is not uncommon for such sensitive microphones to capture sounds that the human ear might not notice in context of a noisy environment, such as an urban street.

Based on the audio recording, the technical markers (see above) strongly suggest that the voice was received and recorded via the camcorder’s microphone.  In fact, the markers are so obvious that the voice appears not to be an EVP (electronic voice phenomenon) but rather an AVP (acoustic voice phenomenon). In other words, as a result of sound pressure waves picked up by the microphone’s capacitor diaphragm or mechanical capsule.

An EVP, by contrast, would appear in a recording not as the result of sound pressure waves entering the microphone, but instead as a result of external EMI (e.g., static) or internal electrical noise (e.g., white noise, hums, buzzes, clicks), or some other discernable electrical activity. In fact, by definition an EVP can only occur when something is found to be intelligible (e.g., a voice, recognizable sound, or utterance) within electrical devices such as internal circuitry. This recording does not demonstrate any sign of being an EVP.

So, to answer the question directly, the only way for the voice not to have come from somewhere in the vicinity of the microphone is for the recording to show signs of being an EVP, which it does not. It is an AVP.

“Some have a hard time hearing “you’re a slave” etc and think it is saying something else. Is there a way to analyse the exact words based on the frequencies, etc?”

Response: See our attempt at isolating the voice in the analysis section above. In our opinion, the mix of voices (i.e., their overlap and intermingling) is too strong to definitively conclude that the voice is saying, “you’re a slave.” Therefore, meaning is inconclusive.

“At about 12 seconds into the video, I believe the voice says “Please stop–” and from there it becomes too difficult to understand. Is there any way to clarify what she’s saying? Also, does it truly end there? Or does it actually continue at a lower tone as we begin to walk?”

Response: Our response here is similar to the previous question. The woman’s voice is too mixed with the other voices to clearly understand what she is saying.  “Please stop” and “don’t hurt me” are certainly the clearest segments of the audio.

“At about 2-3 seconds into the video, there is a slight high pitch sound that I also didn’t notice when recording. To me it sounds like a cat meowing and it occurs right before the voice starts speaking. Is there a way to check the frequencies to see if it is consistent with a meow? If so, that would be even more intriguing considering I’m a huge animal lover who never misses a chance to approach a street cat so I’m wondering if that is also part of the voice.”

Response: Subjectively speaking, to our ears this sound does not resemble anything like a cat’s meow. As discernable from the spectrograph, it appears to range from 700Hz to 7kHz, with at least 10 identifiable harmonic frequencies.  This sound could be from an animal, a human, or even a mechanical device. It’s interesting to note that a typical upper-level fundamental frequency of the human female larynx is around 800Hz.