| |
Abstract:
A comprehensive aeroacoustic approach to speech sound
production is presented. First, a brief discussion of the
theoretical basis for this approach is given. Aeroacoustic theory
is developed as an interaction of two distinguishable components,
namely acoustic (unsteady, compressible, irrotational) and
non-acoustic (incompressible, vortical) air motion. The role of
vorticity as a necessary element for aeroacoustic sources is
highlighted. Second, a conceptual model for speech sound
production based upon these ideas is then presented to frame the
discussion. This conceptual model views speech sound production
as being the acceleration of the acoustic medium by airflow
through a constricted region of the vocal tract. This conceptual
model includes voiced sounds, fricatives, affricates, and stops,
depending on whether the airflow through the constriction is
steady and whether the constriction size varies over times
commensurate with an acoustic time scale. It is shown that the
structure of the airflow is crucial in determining the nature of
the acoustic excitation, and suitable approximations of the
airflow are presented. Finally, the conceptual model is applied
in turn to fricative, stop, voiced, and affricate sound
production by a computer model of air motion in the vocal tract
which in particular highlights the role of vorticity. It is shown
how this approach adds very little computational effort beyond a
traditional lumped-element representation. Spectra and audio
samples of the synthesized speech sounds are presented.
|
