The fine structure of phonation as a biometric.
Sandra Ferrari Disner, Sean A. Fulop, Fang-Ying Hsieh · The Journal of the Acoustical Society of America · 2010
The reassigned spectrogram, an enhanced time-frequency display that has been proposed as a means of focusing in on the details of the phonation process [S. A. Fulop and S. F. Disner, “Advanced time-frequency displays applied to forensic speaker identification,” J. Acoust. Soc. Am. 125, 2530 (2009)] was used to examine a set of sustained vowels produced by 16 speakers of American English. Certain characteristics, such as the relative amplitudes of the formants, the presence or absence of a so-called “voice bar,” and evidence of secondary excitation within a single glottal period, appear to be rather stable across repetitions of the same vowel by the same speaker, but divergent across speakers. The possibility of this method being used to corroborate conventional methods of forensic speaker identification is thus afforded a measure of support.