A comparison of glottal voice source quantification parameters in breathy, normal and pressed phonation of female and male speakers

This study concerns a comparison of different quantification techniques that have been developed to parameterize a voice source that has been estimated using inverse filtering that applies no flow mask (i.e. absolute flow values are not obtained). The speech material consisted of voices of three different phonation types produced by 5 female and 5 male subjects. Quantification of the voice source was computed using three time-based quotients that were extracted from the glottal flow waveforms, two time-based parameters that were extracted from the flow and its first derivative, one amplitude-domain quotient that was defined using both the flow and its derivative, and one frequency domain parameter that was computed from the flow signal. The results showed that phonation types could be separated from each other most effectively when quantification was based on parameters that were extracted between the instant of the maximal glottal opening and the minimum peak of the flow derivative.