5.1 Guan
Xin Guan. (2015). Potential Speaker-Discriminating Power of Speaking Style: Application of Discourse Information Analysis to Forensic Speaker Recognition. International Journal of Law, Language & Discourse 5(1), 38-65.
Abstract: Not only are there differences in the voices that are from different speakers of the same language, but also in the voices that are produced by the same speaker under different conditions or on different occasions. Such nature of human voice makes forensic speaker recognition possible, but difficult. Because in order to link a questioned voice in contact with criminal activity to a known suspect, the forensic speaker recognition expert has to correctly attribute the inevitable differences between two voice samples to either between speaker differences or within speaker differences. The parameters that are currently used in forensic speaker recognition are phonetic features, and their speaker discriminating power has been tested in laboratories with lab recorded audio materials. However, the forensically realistic conditions are far more complex than ideal laboratory conditions. Moreover, forensically realistic conditions have dramatic effects on forensic phonetic parameters. That is, there is a gap between FSR research and practice as far as the efficacy of forensic phonetic parameters is concerned. To bridge the gap the study aims to explore non phonetic features that have the potential to discriminate speakers and at the same time are resistant to within speaker variability in voice and effects of forensically realistic conditions.
Keywords: phonetic parameters, non-phonetic features, speaker-discriminating power, Discourse Information Analysis, individual speaking style
Category: Volume 5
