TUSCALOOSA PARANORMAL RESEARCH GROUP
Because an EVP is by definition an "electronic phenomena," if a vocal approximation is heard without the use of any electronics, even if it is at the time a recording is being made, it is not an EVP. Even if the same sound is heard on the recording device when the audio is reviewed afterwards, it is still not an EVP because it was in its initial form audible without the aid of electronic devices. In these cases, the vocal approximation can be classified as a "disembodied voice," but again, it is not an EVP.
Similarly, a non-vocal sound (percussive, mechanical, structural) that is not heard by investigators during a recording session but is heard on the recording media later in review is also not an EVP. The sound is an audio anomaly, however, and a captured sound does not need to be an EVP to be classified as possibly paranormal. Further review and analysis of the recording is required to make that judgment. But for the purposes here of defining an EVP, the audio anomaly must: 1) be of a vocal nature and 2) must only have been heard in playback of the recording in order to be classified as an EVP.
FURTHER CLASSIFICATION OF EVPs
Once a sound has been identified as an EVP, further classification is useful in acknowledging an EVP's reliability. The fewer technical manipulations and subjective interpretations required to hear and identify the nature of an EVP, the more reliable it is as paranormal evidence data. There are three generally accepted classes of EVP, with Class A being the most reliable and Class C being the least.
The EVP can be heard during simple audio device playback without earphones or the use of audio processing software. Reviewers will also generally agree on what exactly the vocal approximation is with minimal discussion (i.e., words are clearly distinguishable, or sobs/laughter/etc. are clearly of a human vocal nature and cannot readily be attributed to other sources).
The EVP can be heard during audio device playback, but requires headphones to distinguish content, and reviewers may not all initially identify the same content (i.e., words may be muffled or confused; one reviewer may hear crying while another hears laughter, etc.). Class B EVPs will not need audio processing or filtering in order to be identifiable (as soon as any kind of audio processing must be applied to a recorded sound to make it identifiable as an EVP, the classification of that sound becomes a Class C).
The EVP can be heard during audio device playback but requires the use of headphones and amplification and/or additional software processing to be clearly identified. In order for a Class C EVP to remain a Class C rather than a non-EVP audio anomaly, it is best that, after the audio file has been "cleaned up" as much as possible, there remains no question about the content of the file. If reviewers still disagree about the content of what's being heard, the sound can no longer be classified as a reliable EVP and should then be termed an inconclusive audio anomaly of possible vocal nature.