Performance of glottal inverse filtering as tested by aeroelastic modelling of phonation and FE modelling of vocal tract

Performance of glottal inverse filtering (IF) is evaluated in this paper by using speech material produced with a physically-based approach represented with an aeroelastic model of vocal folds and a Finite Element (FE) model of the vocal tract for the vowel /a/. An inverse filtering algorithm was used to estimate the glottal flow from the speech pressure signal generated by the model. Comparison between the estimated glottal flow and the original flow generated by computational modelling shows that the IF method is able to yield an accurate estimate for the glottal flow provided that fundamental frequency of the sound is below 400 Hz.