The effect of sampling variability on overall performance and individual speakers' behaviour in likelihood ratio-based forensic voice comparison.
- Resource Type
- Article
- Authors
- Xiao Wang, Bruce
- Source
- International Journal of Speech, Language & the Law; 2022, Vol. 29 Issue 1, p99-103, 5p
- Subject
- AUTOMATIC speech recognition
BAYESIAN analysis
LINGUISTICS
PHONETICS
MATHEMATICAL variables
- Language
- ISSN
- 17488885
The article discusses the forensic voice comparison (FVC) using Bayesian reasoning and likelihood ratio framework, and the importance of empirically validating system performance. Topics include Linguistic-phonetic variables are used to investigate the effect of varying configurations of statistical model that improve system performance; the testing the performance of automatic speaker recognition; and finding a set of test speakers that produces a validity measure representative for the case.