Open Access
Leonard, Daniel Joseph
Graduate Program:
Aerospace Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
June 25, 2010
Committee Members:
  • Michael H Krane, Thesis Advisor
  • Dennis K Mclaughlin, Thesis Advisor
  • unvoiced
  • turbulent jet
  • vortex
  • speech
  • acoustics
  • aeroacoustics
  • sound
  • source spectrum
This thesis addresses the sound radiated from a turbulent jet-wall interaction in a duct which was measured for several jet-wall interaction geometries. This sound production mechanism is identical to that of unvoiced speech sounds which are relevant to producing consonants and also important for voice quality. Traditionally in these cases, the speech science community has focused on how the vocal tract geometry affects the acoustic lter, and has entirely neglected the effect of the geometry of the aeroacoustic source spectrum. The theory of vortex sound implies that sources of aeroacoustic sound are critically dependent upon the manner in which the jet flows over the wall, in terms of jet/wall geometry, the jet's speed, and especially the structure of the vorticity field. When the local source region aerodynamics, such as the mean jet path relative to the wall and the jet speed are varied, but the acoustic filter is held constant, the differences in the aeroacoustic source are observable in the radiated sound. A combined experimental and analytical study making use of this fact was performed to demonstrate that the aeroacoustic source spectrum for unvoiced speech sounds depends critically on the source region geometry. Such an experiment was performed in a physical model with dimensions commensurate with a life-size human vocal tract. A jet was formed at a 26:1 area constriction in the duct, and passed over a thin tab protruding from the wall and blocking 41% of the duct area. The source region geometry was varied by locating the constriction at different transverse positions, relative to the tab. At low frequencies, this did not change the acoustic response of the duct, but did modify the manner in which the jet flowed over the tab. Measurements of radiated sound performed at four flow speeds for each of three transverse constriction locations show distinct variations in spectral frequency content. The aeroacoustic source spectra were estimated by inverse filtering the radiated sound spectra with a transfer function measured between the source and radiated sound receiver. Since the transfer function remained constant, the trends present in the source spectra matched those observed in the radiated sound. A theoretical estimate of the effect of the jet path on the aeroacoustic source spectrum was made by combining the theory of vortex sound and a simplified jet model. The trends of these source spectral estimates matched those observed in the measured source spectra. All of the distinct differences between each constriction case and flow rate are clearly explained by the theory of vortex sound, whereas the traditional speech science approach, which assumes a flat, bandwidth-limited white noise spectrum would not have been able to predict the presented results. The notion that the turbulent jet's path makes a crucial contribution to the "shape" of the source spectrum was verified , and an increase in the jet's speed displayed an increase in the source spectrum's energy, especially at higher frequencies. It is concluded that unvoiced speech sound production depends as much on the local details of the source region aerodynamics and geometry as it does on the acoustic filter. This result implies that accurate modeling of speech sound production must incorporate details of the source region geometry.