EVP: The Voices That Appear After the Recording Stops
The room was empty. The recorder was running. Nothing audible occurred during the session. When the recording is played back, a voice says something. It is not always clear. It is usually short -- a word, a name, sometimes a phrase. It was not there when the recording was made. Where did it come from? Electronic Voice Phenomena is the name given to this class of audio event, and it has been documented in thousands of recordings by researchers ranging from Latvian psychologists in the 1960s to present-day ghost hunting teams armed with professional audio equipment. The scientific explanation and the paranormal explanations are both specific and neither has fully won the argument.
Key Takeaways
- What EVP Is: Sounds found on electronic recordings that were not perceived during the recording session, subsequently interpreted as voices, names, or paranormal communications. Documented in both analog and digital recording formats across a 60+ year research history.
- Origin: Systematically studied by Latvian psychologist Konstantins Raudive starting in the 1960s. His book Breakthrough (1971) presented over 70,000 recorded "voices" and triggered serious scientific investigation. Friedrich Jurgenson, a Swedish filmmaker, preceded Raudive and is considered the father of EVP research.
- The Science: Primary explanation is auditory pareidolia -- the brain's speech-detection system finding speech patterns in random noise. Secondary explanations include radio frequency interference (especially in older analog recorders), mechanical artifact, stray electrical signals, and in controlled-environment tests, confirmation bias in transcript interpretation (listeners reliably "hear" different things when given different transcripts for the same audio).
- The Anomaly: Some EVP recordings have been analyzed and found to contain phoneme structures and voice-quality characteristics that are more consistent with human speech than with typical audio artifact. This does not verify their paranormal origin but it does complicate simple "it's just noise" dismissals.
Friedrich Jurgenson and the Accidental Discovery
The modern history of EVP research begins with Friedrich Jurgenson, a Swedish documentary filmmaker and ornithologist, who in 1959 recorded birdsong near his summer home and, upon playback, heard what he believed was a male voice discussing bird behavior in Norwegian, followed by a voice he identified as his deceased mother saying his name in German. Jurgenson was not a paranormal researcher; he was, by all accounts, a skeptical professional in the documentary film tradition. The recordings that he subsequently made deliberately, attempting to reproduce and document the phenomenon, form the foundational archive of EVP research.
Jurgenson published his findings in 1964 in Swedish (Röster från rymden, "Voices from the Universe") and attracted the attention of Konstantins Raudive, a Latvian psychologist who had studied under Carl Jung and was working in Germany. Raudive was scientifically trained, methodologically rigorous, and deeply motivated by the death of his wife to investigate whether communication across death was possible. He spent the next several years recording thousands of sessions, documenting over 70,000 recordings that he classified as EVP, and developing a classification system for their clarity and content.
His book Breakthrough, published in English in 1971, was accompanied by a recording that readers could play. The publisher, Colin Smythe, reportedly had the recordings independently tested by audio engineers at Pye Records who confirmed that the sounds were present on the tape and were not explicable by standard recording artifact. This quasi-independent verification was notable at the time, though subsequent critical analysis questioned the engineers' conclusions and methodology. The book generated significant public interest and launched the English-language EVP research community that continues today.
The Scientific Lens: Auditory Pareidolia and the Speech-Detection System
The neuroscience of speech perception is relevant here in ways that are rarely explained in paranormal contexts. The human auditory system is specifically and dramatically tuned for speech detection. This tuning is evolutionary: speech is the primary channel of social information in humans, and the survival advantages of detecting speech -- friend, enemy, danger, food -- were significant enough to produce a detection system that is extremely sensitive, which means sensitive enough to detect speech in conditions where it barely exists. As a side effect, it is sensitive enough to detect speech in conditions where it does not exist at all.
Auditory pareidolia -- the audio equivalent of seeing faces in clouds -- is the standard acoustic name for this phenomenon, and it has been extensively studied. Research by neuroscientist Sophie Scott and others has demonstrated that the human brain activates speech-processing regions in response to acoustic stimuli that contain only the rough statistical signatures of speech (the right frequency distributions, roughly the right temporal structure) without being speech at all. The activation of these systems generates a strong phenomenal impression of hearing speech that the listener cannot easily override through conscious attention.
The critical experimental demonstration for EVP specifically comes from studies in which the same audio recording is presented to listeners with different transcripts: one group is told "you will hear the phrase 'come to me'" and another group is told "you will hear the phrase 'hold the key.'" Both groups reliably hear the phrase they were primed to hear in the same ambiguous audio. This is a dramatic and consistent finding: what people "hear" in ambiguous EVP recordings is not primarily determined by the audio content; it is primarily determined by what they have been told or expect to hear. The implication for EVP research that does not double-blind its transcript methodology is significant.
Class A, B, and C: The EVP Classification System
The EVP research community has developed a classification system to distinguish recordings by clarity. Class A EVP are recordings clear enough to be understood by listeners without transcript, and which multiple listeners independently agree on without prompting. Class B are recordings that require some goodwill but are agreed upon by most listeners given a transcript. Class C are recordings that require transcript guidance and are not agreed upon by independent listeners. The vast majority of EVP recordings fall into Class B and C categories.
The classification system is honest about this distribution: Class A recordings are rare, and within the EVP research tradition they are treated as the highest evidential value. Critics note that even Class A recordings, when subjected to double-blind transcript testing (listeners asked what they hear without being given a transcript first), typically produce significant disagreement. The recordings that achieve cross-listener agreement most reliably are typically the shortest and most common words -- names, yes, no, help -- which are also the words most likely to be found in random noise by a brain looking for speech.
A minority of EVP recordings, when subjected to spectrographic analysis, show frequency and temporal characteristics more consistent with human speech than with audio artifact. These characteristics don't verify paranormal origin but they do mean the original source sound is genuinely ambiguous. The explanations range from distant human voices (the recording equipment picks up sounds at very low level that were present in the environment but inaudible to human ears at the time), to radio frequency transmissions picked up by the recording equipment's circuitry, to the genuinely unexplained. The genuinely unexplained category, in responsible EVP research, is quite small.
Historical Deep Dive: Thomas Edison and the Telephone to the Dead
Thomas Edison, in a 1920 interview with Scientific American, stated that he was working on a device that could be used to communicate with the dead. This statement, from one of the most famous and credentialed scientists in American history, generated enormous public interest and has been cited ever since as evidence that even serious scientific minds took the possibility of post-mortem communication seriously. The statement's context is worth examining: Edison's Scientific American interview was informal and he was known for provocative speculation; it is not clear that he was describing a genuine research project rather than a rhetorical gesture toward the unknown.
The "spirit phone" or "necrophone" that Edison supposedly had in development has never been found in his archived papers or laboratory records. The statement is real. The device is not. But the cultural significance of the statement is independent of whether the device existed: it placed the idea of electronic spirit communication within the frame of legitimate scientific inquiry during a period (the 1920s) when Spiritualism was still a substantial cultural and social movement and when the role of technology in bridging material and spiritual reality was a live question for many educated people.
The post-War I period was a high-water mark for Spiritualist belief in both the US and Europe: the mass deaths of the First World War produced an enormous grief population with intense motivation to find evidence of continued existence. Arthur Conan Doyle, the creator of the hyper-rational Sherlock Holmes, was a committed Spiritualist who toured extensively lecturing on the evidence for life after death. Harry Houdini, his close friend and the period's most famous debunker of fraudulent mediums, devoted the latter part of his life to exposing Spiritualist fraud while maintaining a standing offer of a significant sum to anyone who could demonstrate genuine mediumship under controlled conditions. The EVP tradition that Jurgenson and Raudive systematized in the 1960s drew on a long prior history of technological approaches to spirit communication.
The Skeptic's Corner: Why Ghost Boxes Fail Every Controlled Test
Ghost boxes, Spirit Boxes, and ITC (Instrumental Transcommunication) devices are audio devices that generate white noise or scan AM/FM radio frequencies at speed, producing a stream of audio fragments that EVP researchers interpret for paranormal communication. The rationale is that spirits can modulate the available audio to produce intelligible messages. When these devices are subjected to controlled tests -- double-blind evaluations in which transcripts are assigned randomly or in which the device is asked specific questions whose correct answers are unknown to the researchers -- they fail to produce results better than chance. The information that comes through is consistent with random sampling of radio audio combined with auditory pareidolia applied to the white noise segments.
This failure of controlled testing is the central skeptical argument against all EVP research: phenomena that do not produce results better than chance under controlled conditions are not providing the evidence their researchers believe they are providing, regardless of the phenomenal quality of the experience during uncontrolled investigation. The EVP research community's response is that the controlled testing environment itself prevents phenomena from occurring (spirits communicate more readily in conditions of emotional openness and informal investigation than in clinical conditions), which is an argument that is unfalsifiable and therefore scientifically unprogressive. The debate is currently locked in this epistemological standoff.
Transmission Intercepts: Witness Accounts
"I'm an audio engineer. I've done sessions with paranormal investigators in three different 'haunted' buildings. I have heard things on playback that I cannot immediately explain every time. My explanation is usually 'radio pickup through the pre-amp' or 'HVAC resonance' or 'someone in the parking lot.' But twice I have had recordings that I genuinely did not have a ready explanation for. I keep them. I don't know what to do with them."
-- Listener submission, received January 2026
"After my father died I tried an EVP session in his bedroom with a voice recorder. I heard something that sounded like his name -- my name, said the way he said it. I know what auditory pareidolia is. I know I had a transcript in my head before I pressed play. I also know I cried for twenty minutes. Whatever it was, it helped."
-- Listener submission, received February 2026