Modelling Voiceless Speech Segments by Means of an Additive Procedure Based on the Computation of Formant Sinusoids
Ingo Hertrich, Hermann Ackermann · 2007
A previously developed vowel synthesis algorithm implements formants as sinusoids, amplitude- and phase-modulated by the fundamental frequency (Hertrich and Ackermann, 1999, Journal of the Acoustical Society of America, 106, 2988-2990). The present study extends this approach to the modelling of the acoustic characteristics of aperiodic speech segments. To these ends, a voiceless signal component is generated by adding at each sample point a random parameter onto the formants ' phase progression. Voiceless stop consonants then can be modelled, e.g., by combining a release burst, i.e., an interval in which the formant sinusoids abruptly increase and gradually decrease in amplitude, with formantshaped noise components, representing inter-articulator frication, aspiration, and breathy vowel onset. 1.