A strange sound lasts two seconds. A light crosses one camera. A trigger object activates just after a question. The moment may be compelling—but the evidence does not begin and end with the clip.
Short paranormal videos are easy to share because they remove the waiting, silence and uncertainty around an event. They are also easy to misread for exactly the same reason.
Evidence analysis asks a larger question than “What can we hear or see?” It asks what happened before and after, what every other device recorded, what the environment was doing and what information may have been lost when the moment was edited for presentation.
Protect the original first
The original recording is the closest available record of what the device captured. It should be copied and protected before trimming, amplifying, filtering, resizing or converting anything.
The Scientific Working Group on Digital Evidence recommends preserving data integrity throughout the evidence lifecycle, creating contemporaneous documentation and using working copies and archives. It also describes cryptographic hashes as a way to establish a baseline that can later help verify that a file has not been compromised. Read the SWGDE collection guidance.
A practical evidence package can contain the untouched source file, a protected backup, a separate working copy, the original filename, available metadata, the device used, the date and location, and a record of every change made to a presentation copy.
A hash cannot tell us what caused a sound or image. Its role is narrower: it helps establish whether the file being checked is the same data that was previously hashed.
Rebuild the timeline
A clip beginning half a second before an unexplained knock may hide the footsteps, closing door or question that preceded it. Review should therefore begin with a wider section of the recording, not the shortest dramatic edit.
Note who was in the area, who had just moved, whether anybody entered or left, and whether a known sound was tagged. Check investigation notes and the recordings made immediately before and after the event.
Time displays are useful but not infallible. Devices may be set differently or drift apart. Investigators can align files using shared events—a clap, spoken marker, door closure or other distinctive sound—rather than assuming every displayed timestamp is exact.
Ask what the other devices captured
Multiple recordings do not automatically prove a paranormal cause, but they can clarify whether an event occurred in the environment.
If the same sound appears on recorders in different positions, its relative volume and arrival may help locate the source. If a voice is clear on one device but absent from another, microphone direction, distance, automatic gain and local handling noise become important. If a visual anomaly appears on one camera only, check its lens, infrared illumination, focus, compression and nearby insects or dust.
Video may show an investigator moving when an audio recorder captured a rustle. A second camera may reveal a reflection outside the first camera's frame. A written note may identify a passing vehicle. Corroboration is valuable; contradiction is valuable too.
Separate observation from interpretation
Evidence becomes weaker when a conclusion is built into its description. Compare these two statements:
“A spirit answered the question” versus “A voice-like sound was recorded after the question; its source has not been identified.”
The second statement records what happened without claiming more than the available information supports.
This distinction also matters during group review. Telling listeners what an ambiguous recording supposedly says can influence what they hear. A stronger process presents the unlabelled clip first, collects interpretations independently and compares them afterwards.
Enhancement is not discovery
Amplification, equalisation and noise reduction can make a quiet section easier to hear. They can also change the balance of sounds, emphasise artefacts or encourage the listener to find structure in noise.
SWGDE's current digital-audio guidance recommends conducting examinations on a working copy, documenting relevant decisions and recognising the influence of cognitive bias. It also warns that processing applied before examination may limit later enhancement. Read the audio-enhancement guidance.
Every public enhancement should therefore be labelled. Keep the original available, state what processing was applied and avoid presenting an altered version as though it came directly from the recorder.
Test the ordinary explanations
A useful review actively searches for ways the event could have occurred without a paranormal cause. That means checking human speech, movement, clothing, equipment handling, radio interference, plumbing, wildlife, traffic, airflow, vibration and reflections where relevant.
Recreating an effect does not prove that the original event had the same cause. It shows that a credible mechanism exists and must be considered.
Equally, failing to recreate an event does not make it paranormal. Conditions may have changed, a source may no longer be present or the available information may simply be incomplete.
Use conclusions that match the evidence
Evidence analysis does not have to end with “real” or “fake”. More honest outcomes include explained, likely contamination, ambiguous, unexplained and potentially significant.
The SWGDE guidance for digital-audio authentication says conclusions should be supported by the analysis, should not be overstated and should avoid claims of absolute certainty. See the authentication guidance.
“Unexplained” is not a synonym for “paranormal”. It means the information available has not produced a reliable explanation. “Ambiguous” is not a disappointing result. It is an accurate description when a recording supports more than one reasonable interpretation.
Show enough for others to question it
Responsible public evidence should give viewers room to reach their own conclusion. Where privacy and venue agreements allow, provide the unlabelled clip, the wider context, who was present, whether the event was heard live, what other devices captured and which ordinary explanations were considered.
The strongest presentation is not always the most dramatic. It is the one that allows the audience to understand what is known, what has been tested and what remains uncertain.
Preserve the file. Rebuild the moment. Test the alternatives. Report only what the evidence can carry.
Expect Nothing. Question Everything.