If a suspect and a criminal are the same person, how likely is the similarity between two voice recordings?
If they are not the same person, how likely would be the similarity?
The ratio of these two probabilities is called the likelihood ratio, or strength of evidence. The higher the strength of evidence, the stronger the evidence.
Due to the unique structure and characteristics of human voice it is possible to distinguish voices audibly as well as visually through the use of spectrographs.
Amongst the most common outcomes from a voice identification process are: possible identification, probable identification, possible elimination, probable elimination, and inconclusive.
Specific guidelines and standards are followed to ensure that the conclusions are consistent and legally-valid. There is a minimum number of words that must be present in a recording in order for this task to be performed according to the standards.
The forensic comparison of voice samples is an extremely complex process requiring expert knowledge in not just one, but in several different specialty areas related to speech science.
In its most common form speaker identification involves the comparison of one or more samples of an offender’s voice and is compared to one or more samples of the suspect’s voice. High quality exemplars are made which are critical in allowing an accurate comparison with unknown voice samples.
A preliminary forensic audio examination must be performed in case the unknown and known voice samples do not meet specific guidelines to allow continuation of the examination.