A new software for forensic voice comparison: IMPAVIDO

IMPAVIDO (Integrated Methods for Parametric Voice Identification) is a new tool for forensic voice comparison.

Forensic speaker recognition: an introduction (About page)

Ideal for integration into a semi-automatic acoustic-phonetic method, it implements a statistical comparison of spectroacoustic features (fundamental frequency and formant frequencies of Italian vowels), using two different models integrated together: parametric models and non-parametric models.

The parametric model implemented in IMPAVIDO (called eIDEM) is inspired by the SPREAD module of the IDEM software (IDEntification Method), traditionally used by the Carabinieri Corps.

The implementation of the non-parametric model (eSMART) is inspired by the SMART software, traditionally used by the Italian Police.

The software uses a methodology based on the likelihood ratio (Bayesian Interpretation Framework), in line with ENFSI recommendations.

Screenshots

Main graphical interface of IMPAVIDO. Univariate Gaussian probability densities.

IMPAVIDO - Confidence ellipses

IMPAVIDO - Three-dimensional plot of bivariate Gaussian probability densities.

IMPAVIDO - Confidence ellipse chart for all vowels in the casework.

IMPAVIDO - Comparison output chart with eSMART (non-parametric model).

Please use the following email address:

impavido@biometriaforense.it