Forensic speech analysis covers a number of areas including identifying whether a recording has been tampered with, deciphering unclear speech, unravelling speech codes and designing voice line-ups.
One of the most common enquiries is the analysis of recordings to help identify the speaker. Samples of known and unknown speech are subjected to a fine-grained phonetic analysis.
Forensic speech analysis comprises a number of subfields, including audio authentication (whether a recording has been tampered with), speech decoding (deciphering unclear speech), speech decrypting (unravelling speech codes) and designing voice line-ups (akin to visual line-ups). A common question in the analysis of evidential recordings is speaker identification. The most reliable method used to perform such analysis is using a phonetic-acoustic approach. A speaker is identified (on a verbal – rather than numerical - scale of judgments) by systematic comparison with a ‘known’ or ‘reference’ recording of a suspect/defendant.