Modeling the source for fricative consonants
Christine H. Shadle · The Journal of the Acoustical Society of America · 1984
Turbulence noise in the vocal tract, which plays a role in the production of fricative and stop consonants, is usually modeled by placing one or more sound-pressure sources with smooth spectra in series with a one-dimensional model of the vocal tract [G. Fant, Acoustic Theory of Speech Production (Mouton, 1960)]. In this paper it is shown that this model is consistent with the sound generated when a jet of air impinges on an obstacle. Theoretically, the sources at the obstacle are dipoles, and thus are equivalent to a pressure source. Experimentally, the power spectrum of the sound generated by a mechanical model corresponds closely to the transfer function simulated for a tube with identical area function and a pressure source at the location of the obstacle. When no obstacle is present downstream of the constriction in the model, no such correspondence is observed, indicating that for this case a distributed source is a better model of the sound generation mechanism. Analysis of spoken sustained fricatives shows a similar division between stridents (with obstacle) and non-stridents (without an obstacle). [Work supported by NIH.]