Identifying unfamiliar voices: Examining the system variables of sample duration and parade size

Abstract

Voice identification parades can be unreliable due to the error-prone nature of earwitness responses. UK government guidelines recommend that voice parades should have nine voices, each played for 60 seconds. This makes parades resource-consuming to construct. In the present paper we conducted two experiments to see if voice parade procedures could be simplified. In Experiment 1 (N=271, 135F), we investigated if reducing the duration of the voice samples on a nine-voice parade would negatively affect identification performance using both conventional logistic and signal detection approaches. In Experiment 2 (N=270, 136F), we first explored if the same sample duration conditions used in Experiment 1 would lead to different outcomes if we reduced the parade size to include only six voices. Following this, we pooled the data from both experiments to investigate the influence of target-position effects. The results show that 15s sample durations result in statistically equivalent voice identification performance to the longer 60s sample durations, but that the 30s sample duration suffers in terms of overall signal sensitivity. This pattern of results was replicated using both a nine- and a six-voice parade. Performance on target-absent parades were at chance-levels in both parade sizes and response criteria were mostly liberal. Additionally, unwanted position effects were present. The results provide initial evidence that the sample duration used in a voice parade may be reduced, but we argue that the guidelines recommending a parade with nine-voices should be maintained to provide additional protection for a potentially innocent suspect given the low target-absent accuracy.

Time to reflect on voice parades: The influence of reflection and retention interval duration on earwitness performance

“…Although the Home Office recommend that voice samples should be 60s, Pautz et al (2023) showed that there was no difference in performance between 15 s and 60 s samples, thus for the present research, excerpts featuring the interviewees were spliced together to produce a 15 s sample for each speaker. The voice samples were all taken from different sections of each interview so that the content of speech differed across speakers.…”

“…A total of N = 180 participants were recruited via the online recruitment platform Prolific.co. The sample size, per condition, is either similar to or exceeds contemporary studies in the area (Kerstholt et al, 2006;Pautz et al, 2023;Perfect et al, 2002;Philippon et al, 2013;Smith et al, 2020). The present research, like most voice identification research, was constrained by the cost of participant recruitment.…”

“…The voice parade design used in the present set of experiments was based on the same rigorous procedure used in Pautz et al (2023), which was modelled on the recommendations for constructing a voice parade set by the Home Office (2003) in England and Wales. Specifically, a total of nine voice samples were included in the parade which were presented serially.…”

“…Voice identification procedures are conducted across the world, including in Britain, Australia, Canada, various European states, and the United States (Broeders & van Amelsvoort, 2001; Cantone, 2010; de Jong‐Lendle et al, 2015; Gfroerer & Jessen, 2022; Laub et al, 2013; McGorrery & McMahon, 2017). While the utility of voice parades in certain crime situations is undeniable and has, for example, been crucial in the successful prosecution of suspects in numerous British legal cases (Clifford, 1983; see Nolan, 2003, for an overview of R v. Khan and Bains, a case involving arson by murder), those familiar with voice identification research will be well aware that the outcomes of lab‐based studies generally show that accuracy rates are low (e.g., Kerstholt et al, 2004, 2006; Pautz et al, 2023; Smith et al, 2020). Error‐rates are particularly high in target‐absent parades, which simulate an innocent suspect having been apprehended (e.g., Smith et al, 2020, 2022).…”

“…Alternatively, they stay in the lab and complete a filler task, 1 such as a wordsearch (Smith et al, 2022) or an unrelated questionnaire (Philippon et al, 2007). Filler tasks are shorter in duration, generally lasting between 5 (McAllister et al, 1993; Pautz et al, 2023; Smith et al, 2020) and 30 min (Philippon et al, 2007) and serve the purpose of ensuring the encoded voice is stored in long‐term, rather than short‐term memory, while preventing active rehearsal (Hartshorne & Makovski, 2019). Finally, during the ‘ test phase ’ participants listen to the voice included in the parade and make a decision about whether the target voice was present or absent.…”