Phonological content impact on wrongful convictions in Forensic Voice Comparison context

Abstract

Forensic Voice Comparison (FVC) is increasingly using the likelihood ratio (LR) in order to indicate whether the evidence supports the prosecution (same-speaker) or defender (different-speakers) hypotheses. Nevertheless, the LR accepts some practical limitations due both to its estimation process itself and to a lack of knowledge about the reliability of this (practical) estimation process. It is particularly true when FVC is considered using Automatic Speaker Recognition (ASR) systems. Indeed, in the LR estimation performed by ASR systems, different factors are not considered such as speaker intrinsic characteristics, denoted "speaker factor", the amount of information involved in the comparison as well as the phonological content and so on. This article focuses on the impact of phonological content on FVC involving two different speakers and more precisely the potential implication of a specific phonemic category on wrongful conviction cases (innocents are send behind bars). We show that even though the vast majority of speaker pairs (more than 90%) are well discriminated, few pairs are difficult to distinguish. For the "best" discriminated pairs, all the phonemic content play a positive role in speaker discrimination while for the "worst" pairs, it appears that nasals have a negative effect and lead to a confusion between speakers.

Voice Comparison and Rhythm: Behavioral Differences between Target and Non-target Comparisons

“…Automatic Speaker Recognition (ASpR) is considered as one of the most appropriate solution [3]. Even if impressive low error rates (≈ 1% [4,5]) were reported in the last years, the forensic scenario is still very challenging for ASpR for several reasons [6].…”

“…Second, the speaker himself is an important factor [10,6,11]. A speaker could be ill, or under the influence of stress, alcohol or other drugs.…”

“…A speaker could be ill, or under the influence of stress, alcohol or other drugs. The social and linguistic environment of the unknown speaker is unknown by construction.Speaker's intrinsic characteristics may have a huge impact on the intraspeaker variability [6,11]. Indeed, [11] showed that speakers do not behave the same way in response of similar conditions: some speakers will be quite robust with limited LR variation when some other are showing a huge variation.…”

Impact of rhythm on forensic voice comparison reliability

“…The social and linguistic environment of the unknown speaker is unknown by construction (so, for example, an unknown native or second language should be taken into account by the forensic experts). The speakers intrinsic characteristics may have a huge impact on the intraspeaker variability and therefore put into question the validity and reliability of FVC [8,19,20]. Indeed, in [19], we showed that speakers do not behave the same way in response of similar condition changes: some speakers will be quite robust with limited LR variation when some other are showing a huge variation.…”

“…And finally, the speaker himself is an important factor [18,8,19,20]. A speaker could be ill, or under the influence of stress, alcohol or other factors.…”