EVP analysis means reviewing a recording that contains an apparently unexplained, voice-like sound. Start with the untouched file, document what you hear, compare ordinary explanations, and make any processing reproducible. An unclear sound, a clean spectrogram, or agreement between listeners does not establish a paranormal source.
1. Preserve the original recording
Copy the recorder’s native file directly to your archive and keep a separate working copy. Record the file name, device, session date, time zone, location and who handled it. Keep a second backup before clearing the recorder.
If the source is MP3, preserve that MP3. Converting it to WAV does not restore discarded information. Use lossless working exports to avoid adding repeated lossy compression, and record any conversion. This follows the preservation principle in SWGDE’s forensic audio guidance; a hobby workflow is not a professional forensic examination.
A SHA-256 file hash can help check whether a copy matches the archived file. It does not establish when the sound was recorded, whether the original was edited, or what produced the sound. In Windows PowerShell, Get-FileHash -Algorithm SHA256 -LiteralPath 'C:\Recordings\session.wav' calculates a hash; replace the example path with your file and log the result.
2. Listen before filtering or suggesting words
- Listen at a comfortable level to the original in context, including the surrounding seconds.
- Write the start and end time of an interesting sound and a neutral description, such as “brief two-syllable sound.”
- Compare the session log for footsteps, clothing, conversations, radios, plumbing and handling noise.
- Ask independent listeners what they hear without supplying a transcript, a ghost story or an expected answer. Keep disagreements and “no words heard” in the record.
Research has examined expectation and interpretation in EVP listening, with mixed findings across experiments. A 2022 PLOS ONE study is a useful reason to take listener disagreement seriously. A phrase that seems to answer a question can still be coincidental or misheard.
3. Use visual tools to investigate ordinary causes
A waveform helps locate sudden peaks and interruptions. A spectrogram displays frequency content over time. Neither is a paranormal detector: recorded radio speech is still speech, and visual resemblance to a voice cannot identify its source.
Compare suspected noises with a new control recording of the same device and room. Test one possible source at a time, such as a strap touching the microphone. Record test conditions and differences. Reproducing a noise can support an ordinary explanation; failing to reproduce it leaves the cause unresolved.
4. Process a copy and document every change
There is no universal filter setting that reveals a hidden message. Try the smallest change needed for a specific problem, save a separately named version, and compare it with the original at a similar listening level. Record the application, version, effect and parameter values.
Audacity’s Noise Reduction documentation explains the need for an appropriate noise-only profile and warns that reduction can damage wanted audio or introduce artifacts. Do not keep increasing processing until a preferred word appears. AI isolation and automatic transcripts are outputs to question, not independent witnesses.
| Log field | What to record |
|---|---|
| Source | Original file name, format and archive location |
| Region | Start/end time and surrounding context |
| Observation | Neutral description and separate listener notes |
| Processing | Software, settings and output file name |
| Conclusion | Explanation tested, result and remaining uncertainty |
Report the limits alongside the clip
Share an unprocessed excerpt with context alongside any enhanced version, clearly labeled. Describe missing information, competing explanations and editing steps. Avoid labels such as “authenticated spirit voice,” “scientifically proven” or “court-ready.” A hash confirms file integrity relative to a reference; it does not prove paranormal origin.
The illustration below is a prompt to compare possible sound sources, not a diagnostic test for genuine paranormal signals.
Choose a recorder for documentation, not promises
Check the current manual for supported file formats, microphone inputs, monitoring and transfer options. Record a test before a session. No recorder can identify a sound as a spirit.
Affiliate disclosure: HauntGears may earn a commission from qualifying purchases through its product links, at no additional cost to you. Compare the Olympus WS-852 listing and the Zoom H1essential bundle listing; verify the seller’s current contents and specifications before buying.
For preparation, use the EVP session checklist. For terminology, read what an EVP recorder means.
Common questions
Can software prove an EVP is paranormal?
No. Software can display and process recorded audio. A source that remains unidentified should be reported as unidentified.
Should I delete clips that have an ordinary explanation?
Keep them with your session records when practical. They provide useful reference sounds and show how your interpretation changed.
Does a clearer enhanced clip mean a better finding?
Not necessarily. Processing changes the signal and can change what listeners perceive. Preserve the original and explain each alteration.
Discover more from Haunt Gears
Subscribe to get the latest posts sent to your email.
