How to Record EVPs (Electronic Voice Phenomena): A Step-by-Step Guide

Some links in this article are affiliate links — if you buy through them we may earn a small commission at no extra cost to you. This article documents reported phenomena and does not assert supernatural causation.

Electronic Voice Phenomena — sounds interpreted as anomalous voices or utterances captured on audio recording equipment — have been a fixture of paranormal investigation since the 1950s, when Friedrich Jürgenson, a Swedish film producer, reported hearing unexpected voices while playing back birdsong recordings. Decades later, researchers and amateur investigators continue to bring recorders into reportedly active locations, hoping to capture something they can’t easily explain. Whatever your position on the underlying cause, the recording methodology matters enormously. Poor technique produces false positives at an alarming rate. This guide walks through equipment selection, site preparation, recording discipline, and review workflow — with the goal of producing audio that can survive serious scrutiny.

Choosing the Right Recorder

The recorder is the foundation of any EVP session. The debate among investigators often centers on whether dedicated digital recorders or analog tape machines are preferable. Proponents of tape — a position associated with early researchers like Konstantīns Raudive, who documented thousands of recordings in his 1971 book Breakthrough — argue that the electromagnetic characteristics of magnetic tape create conditions amenable to anomalous capture. The majority of contemporary investigators, however, work digitally, citing the cleaner noise floor, the ease of review, and the accessibility of editing software.

For digital recording, three recorders consistently appear at the top of field recommendations across the investigation community:

• Zoom H1n: Compact, rugged, and capable of recording at 24-bit/96kHz — a sample rate well above what most audio anomalies would require. Its X/Y stereo microphone configuration helps with directionality during review. The Zoom H1n is the most commonly cited mid-range choice among field investigators.
• Tascam DR-05X: A strong alternative with a similarly low self-noise floor and a slightly more ergonomic form factor for solo investigators. Its built-in limiter helps prevent clipping during sudden environmental sounds. Find the Tascam DR-05X if the Zoom is unavailable or out of budget.
• Sony ICD-PX370: The budget entry point. At roughly a third of the price of the Zoom, it sacrifices some dynamic range and records at 16-bit rather than 24-bit. For investigators just starting out or working multiple rooms simultaneously, the Sony ICD-PX370 is a practical choice — though its onboard microphones introduce more self-noise at high gain settings.

Regardless of which recorder you choose, record at the highest quality setting the device allows. Storage is cheap. Audio resolution is not recoverable after the fact.

If your budget allows for a separate external microphone — rather than relying on built-in capsules — a small-diaphragm condenser on a short boom arm can reduce handling noise significantly. Recorders with an external mic input (the Zoom H1n and Tascam DR-05X both have 3.5mm TRS inputs) can accept a dedicated field recording microphone, which keeps the capsule away from the body of the recorder and the investigator’s hands.

Location Selection and Baseline Testing

No piece of equipment compensates for a poorly chosen or poorly understood recording environment. Before any formal session begins, the location must be assessed for ambient noise sources — methodically, not casually.

Walk every room you intend to use and listen for:

• HVAC systems, including ductwork that pops or hisses when temperature changes
• Refrigerators, ice makers, and water heaters with cycling compressors
• Plumbing — pipes that knock, drip, or carry sound from adjacent rooms
• Distant traffic, rail lines, or aircraft flight paths that produce low-frequency rumble at regular intervals
• Loose windows, settling structures, or doors that move in response to pressure differentials
• Electronic interference: fluorescent ballasts, dimmer switches, Wi-Fi routers, and unshielded power cables can all introduce artifacts into the audio chain

Once you have walked the space, record a baseline pass: place the recorder in the intended session location, press record, and leave the room completely for at least fifteen minutes. Review that audio before the session begins. What you hear on that baseline recording — clicks, hum, intermittent sounds — is your control track. Any sound that appears during the session must be compared against the baseline before it is labeled anomalous.

If the space has an active HVAC system that cannot be disabled, note the cycling intervals. A compressor that kicks on every eight minutes will produce a low rumble that can, under pareidolia conditions (more on that below), be interpreted as a voice. Document it. Timestamp it on your baseline track. Remove it from consideration during review.

Recording Technique: What to Do and Why

Good recording technique is largely about discipline — reducing the number of variables that can corrupt the audio. These habits should be consistent on every session, regardless of location or experience level.

Announce Time, Date, and Location

At the start of every recording, speak clearly into the recorder: the date, the time, the specific location within the building, and who is present. This is not a ritual — it is documentation. It ensures that if an audio file is ever separated from its metadata, the content is still identifiable. It also establishes the voices of every person in the room on record, creating a reference library for voice comparison during review.

Ask Questions With Deliberate Pauses

The standard EVP session protocol involves asking questions aloud and leaving silence afterward — typically 20 to 30 seconds of quiet — before speaking again. This pause is where any anomalous audio would theoretically be captured. It also creates clean segments during review, making it straightforward to isolate the gaps and listen to them independently.

Keep questions simple and direct. Compound questions (“Can you tell us your name and when you died?”) produce ambiguous response windows. Single, specific prompts (“What is your name?” — pause — “How old are you?” — pause) keep the structure clean. Note each question and its timestamp in your field log. This becomes essential during audio review.

Physically Isolate the Recorder From Investigators

This is the single most violated rule in amateur EVP work, and it produces more false positives than any other factor. When investigators hold the recorder, every joint crack, swallow, breath, and clothing rustle couples directly through the casing into the capsule. The solution is simple: place the recorder on a stable, non-resonant surface — a small sandbag, a folded cloth, a purpose-built microphone stand — and step away from it. At minimum, maintain a distance of six feet between the nearest investigator and the recorder during question pauses.

Require all participants to remain still and hold their breath during the pause window. Brief team members before the session: even a whispered aside between investigators, inaudible to the human ear in a large room, can register clearly on a sensitive condenser capsule.

Verbally Tag Every Noise You Hear

When a known noise occurs during a session — someone shifts position, a floorboard settles, a car passes outside — say so immediately and clearly, on the recording: “That was a car outside,” or “I just adjusted my stance.” These in-session tags are invaluable during review. An untagged creak that sounds anomalous two weeks later becomes much less interesting when the session notes confirm that someone walked across the floor at that exact timestamp.

Understanding the EVP Classification System

The most widely used taxonomy for categorizing EVP recordings comes from the American Association of Electronic Voice Phenomena (AA-EVP), the organization founded by researchers Sarah Estep and, later, Tom and Lisa Butler. Their three-tier classification system gives investigators a shared vocabulary for describing what they’ve captured:

• Class A: A voice or sound clearly audible to the unaided ear, without headphones, and generally agreed upon by multiple listeners without coaching or suggestion. Class A captures are rare. When they do occur, they warrant the highest level of scrutiny before any anomalous claim is made.
• Class B: Audible with headphones or modest amplification, and recognizable to most listeners as voice-like, though individual interpretation of the specific words may vary. The majority of reported EVPs fall into this category.
• Class C: Faint, barely audible, and subject to significant disagreement between listeners about whether a voice is present at all. Class C recordings are the most vulnerable to pareidolia — the cognitive tendency to find patterns, including voices, in random noise — and should be treated with corresponding caution.

When presenting a recording to others for evaluation, apply one important discipline: do not tell the listener what you think you hear before they have listened independently. Suggestion dramatically skews interpretation. Play the clip without commentary, ask for a written response, and compare results across multiple listeners. This is the closest thing to a blinded test that field conditions allow.

The Review Process: Audacity Workflow and Headphone Discipline

Audio review is where sessions are won or lost. Rushing it, or doing it in a distracting environment, is how legitimate noise gets promoted to EVP status — and how genuine anomalies get missed.

Setting Up Audacity for Review

Audacity is the standard free, open-source audio editing platform used across the paranormal investigation community. Import your session file and your baseline file as separate tracks. Listen to them in parallel at first, using the baseline as a reference against which to compare sounds in the session track.

For the session review, use the spectrogram view (View → Spectrogram) alongside the standard waveform. Speech and voice-like sounds appear in roughly the 300Hz–3,000Hz range. If something appears anomalous on the waveform but falls entirely outside that band, it is less likely to be voice and more likely to be mechanical or electronic interference.

Apply minimal processing during initial review. If a sound is only audible after aggressive amplification, pitch-shifting, or noise reduction, document that fact — it belongs in the Class C category and the processing chain must be disclosed when sharing the recording. Do not apply heavy noise reduction and then present the output as a clean capture; that process introduces its own artifacts, which can mimic voice-like characteristics.

Headphone Selection Matters

Do not review EVP audio through laptop speakers or earbuds. Low-frequency room noise and voice-range artifacts require a full-range, accurate headphone to assess properly. A closed-back pair with a flat frequency response — studio monitor headphones rather than consumer bass-boosted models — gives you the most honest representation of what the recorder actually captured. A reliable pair of closed-back studio monitor headphones is one of the most cost-effective investments an investigator can make.

Review in a quiet room. Not a room with the television on in the background. Not with music playing. Quiet, with the lights on, at a reasonable hour. Fatigue is a significant contributor to false positives — the tired brain finds patterns aggressively. If you have been on-site for six hours and it is 3 a.m., the audio review belongs to the following day.

Common False Positives: A Field Reference

The following are the most frequently misidentified sounds in EVP recordings. Knowing them on sight — or rather, on sound — prevents a significant proportion of false reports:

• HVAC cycling: Produces a low-frequency rumble with a soft onset and offset. Can sound like a distant voice or a moan when heard in brief clips without context. Baseline recording catches this definitively.
• Refrigerator or freezer compressor: The startup transient can produce a click or pop followed by a sustained hum. The harmonic content of compressor hum can, under pareidolia conditions, suggest rhythmic speech.
• Investigator breath: Even at a distance of six feet, a sharp exhale or inhale can register on a sensitive capsule. At high gain settings, breath sounds can be mistaken for whispered words. Tag all breathing-related sounds during the session.
• Clothing rustle: Synthetic fabrics in particular produce a wide-band noise that can suggest sibilance — the “s” sounds in speech — when amplified. Investigators in session should wear natural-fiber layers and move as little as possible during question pauses.
• Distant traffic and low-frequency rumble: Ground-conducted vibration from passing trucks or trains produces infrasound and low-frequency content that recorder capsules pick up even when the human ear does not consciously register it. Cross-reference with baseline and local traffic patterns.
• Recorder self-noise and digital artifacts: Budget recorders at maximum gain produce a noise floor that includes random broadband hiss. Within that hiss, the pattern-seeking brain frequently finds voices. This is the most common source of Class C false positives.
• Electromagnetic interference: Unshielded cables near power sources, fluorescent light ballasts, and nearby radio transmitters can induce audible artifacts directly into the recording chain. Use shielded cables where possible and keep the recorder away from power strips and mains wiring.

Common Mistakes

Even experienced investigators fall into these habits. A brief checklist worth reviewing before every session:

• Skipping the baseline recording because the location “seems quiet”
• Holding the recorder during question pauses
• Failing to verbally tag known noises in real time
• Reviewing audio immediately after a long on-site investigation, while fatigued
• Playing a clip for other investigators and saying “I think I hear the word ‘help’ — do you hear it?” before they have listened independently
• Presenting a heavily processed audio clip without disclosing the processing chain
• Recording in mono when the equipment supports stereo — stereo provides directional information that helps distinguish between sounds originating from different parts of the room

Electronic Voice Phenomena research occupies an unusual position: it is taken seriously enough to have produced peer-reviewed acoustic studies, attracted attention from organizations like the Institut Métapsychique in Paris, and generated a substantial body of documented cases. It is also an area where confirmation bias, auditory pareidolia, and poor recording technique produce false positives at a rate that has frustrated researchers for decades. The answer to that tension is not to abandon the inquiry — it is to raise the methodological floor. Quiet rooms, quality recorders, disciplined technique, and honest review are not obstacles to finding anomalous audio. They are the conditions under which anomalous audio, if it exists, might actually be found.

Related case file: Best EMF Meters for Paranormal Investigation in 2026.

Related case files

• Best EMF Meters for Paranormal Investigation in 2026
• How to Set Up Your First Paranormal Investigation Kit (Under $300)
• The Enfield Poltergeist: 18 Months of Reported Paranormal Activity in London
• Ed and Lorraine Warren: A Career in Demonic Cases (and the Conjuring Universe)
• Best Night Vision Monoculars for Cryptid Hunting and Skywatching