Forensics/Analysis/Example 2

Selection of samples

Vowel segment with noise

analysis of glottal source

The audio sample corresponds to a telephone recording that picks up the conversation between two adult individuals, male, identified as 1 speaker and speaker 2 . Like it is a telephone channel recording, recording acoustic characteristics and the intelligibility of the message is limited by the bandwidth of the channel itself, ambient noise and proximity of the partners to the item pickup (microphone). There have been a total of 9 segments of voice (lengthening of vowels /a/ and/e /) samples of certain voice and dubitada for the analysis of phonation and biometric parameterization of vocal cord. Each segment has been previously edited, and isolated from the rest to further biometric analysis with the BioMet@Fore tool v 3.1

1. The audio sample has been digitally processed to improve its quality mitigating and minimizing noise fragments that could mask the interventions of both speakers.
2. Each segment of voice is subjected to the BioMet@Fore tool v 3.1 for the evaluation of biometric profile of the speaker, including parameters of distortion, cepstral, glottal source and biomechanics of the vocal fold, etc.
3. The samples will be part of the model of suspect to be collated the standardized population model and questioned voice.

The diagram below visually shows that the joint dubitado (Questioned-T) is located next to the model(Suspect) and that both are far from the standard set of population (line-up).

The result supports the hypothesis of the prosecutor referred to the likelihood that the sample dubitada and indubitable come from the same speaker.