Digital Voice Recorders for EVP Work: A Practical Guide

## Electronic voice phenomena — a brief context Electronic voice phenomena (EVP) refers to sounds or speech-like patterns found in audio recordings that were not audibly present at the time of recording. The term was popularised in the 1960s and remains one of the most discussed — and contested — areas of paranormal investigation. The critical issue is that the human auditory system is exceptionally good at detecting speech patterns in noise (a well-documented cognitive tendency called apophenia or pareidolia in the auditory domain). Any EVP methodology must account for this rigorously. ## What type of recorder to use **Digital recorders with an internal microphone** are the standard tool. Look for: - A **low noise floor** — stated in dBu or dBFS. A lower number (e.g., −80 dBu or better) means less inherent self-noise and thus less for pareidolia to work on. - **Uncompressed or lossless format** (PCM WAV at 16-bit/44.1 kHz minimum). Compressed formats such as MP3 introduce compression artefacts — block noise, pre-echo — that can be mistaken for speech. This is not speculation; compression artefacts are a well-documented source of false EVP. - **Manual gain control** — automatic gain control (AGC) will pump and breathe in quiet environments, creating amplitude modulation that the ear reads as "voice-shaped." Disable AGC for EVP work. - **Mono or stereo** — stereo allows directional cross-referencing (if a sound appears in both channels at similar amplitude, the source is roughly equidistant from both capsules, which can help locate it). ## Pre-session preparation Record a room-tone reference before the investigation begins and after it ends. This gives you a known-clean baseline to subtract spectrally and helps identify HVAC noise, distant traffic, or other environmental floor. Test your recorder by speaking known words at known distances so you understand how your device renders real speech. This calibrates your ear for that recorder's characteristics before you review investigation audio. ## Review protocol matters more than the device Double-blind review — where the reviewer does not know what to expect or at what timestamps — produces far fewer confirmed EVP than non-blind review. Consider having a second, uninvolved listener review audio without being told what they are listening for. Keep a session log noting every sound made by investigators: footsteps, clothing movement, whispers, cleared throats, external sources (traffic outside, adjacent room sounds). These are your null-hypothesis candidates. ## What to avoid - Voice-activated (VOX) recording: the activation transient and the gap at the start of each activation clip are artefact-prone zones. - Running the recorder in a pocket or bag where clothing friction generates broadband noise. - Cheap recorders with high self-noise — their thermal noise floor creates an ideal substrate for hearing things that are not there. ## Summary The recorder is only a data capture device. The methodology — session logs, blind review, noise floor documentation, and honest consideration of mundane explanations — determines whether your results are meaningful.