This study evaluates methods that capture time-dynamic properties of diphthongs produced by speakers of Viennese German for application in a forensic setting. Polynomials, discrete cosine transform and B-splines along with experimental features based on bent-cable regression models were used to characterise the first three formant tracks of two /aE/ diphthong segments. The resulting coefficients were in turn used as parameters in a speaker discrimination procedure based on likelihood ratios which were calculated using a multi-variate kernel density formula (MVKD). A comparison of the achieved performance based on cross-validation is presented in terms of equal error rate (EER) and the log-likelihood ratio cost metric as well as DET plots.
Author:
Enzinger, Ewald
Affiliation:
Acoustics Research Institute, Austrian Academy of Sciences, Vienna, Austria
AES Conference:
39th International Conference: Audio Forensics: Practices and Challenges (June 2010)
Paper Number:
2-2
Publication Date:
June 17, 2010
Subject:
Speech and Forensics - Voice Identification
Click to purchase paper as a non-member or you can login as an AES member to see more options.
No AES members have commented on this paper yet.
To be notified of new comments on this paper you can subscribe to this RSS feed. Forum users should login to see additional options.
If you are not yet an AES member and have something important to say about this paper then we urge you to join the AES today and make your voice heard. You can join online today by clicking here.