Speaker Identification Using Laughter in a Close Social Network

Forensically-relevant research on laughter is extremely limited in the literature, however, experts have reported analysing laughter in forensic speaker comparison casework (Gold and French 2011). This paper describes a preliminary investigation into the potential speaker-specificity of laughter. A close social network of 7 undergraduate university students took part in an open speaker identification task containing 4-second samples of their laughter. Overall, the network members performed much worse than in a similar study using speech samples (see Foulkes and Barron 2000), as each network member identified only one speaker correctly. The largest number of correct identifications of any speaker was three, while another three of the network members were never correctly identified. Previous studies that have also investigated laughter using voice line-ups have reported higher identification rates (Philippon et al. 2013; Yarmey 2004). The differences between the results of the present study and previous studies may be explained by qualitative and quantitative differences in the laughter samples used, particularly differences in voicing and sample length. This suggests that longer samples of specifically voiced laughter may facilitate higher naïve speaker identification rates. Further research is still needed on the possible speaker-specificity of voiced laughter but it may have the potential to be developed for use as a speaker discriminant in forensic phonetic casework.