Sound transmission in strongly-curved slowly-varying cylindrical and annular lined ducts with flow

In this paper we consider the propagation of acoustic waves along a curved hollow or annular duct with lined walls. The curvature of the duct centreline and the wall radii vary slowly along the duct, allowing application of an asymptotic multiple scales analysis. This generalises Rienstra's analysis of a straight duct of varying cross-sectional radius. The result of the analysis is that the modal wavenumbers and mode shapes are determined locally as modes of a torus with the same local curvature, while the amplitude of the modes evolves as the mode propagates along the duct. The duct modes are found numerically at each axial location using a pseudo-spectral method. Unlike the case of a straight duct, there is a fundamental asymmetry between upstream and downstream propagating modes, with some mode shapes tending to be concentrated on either the inside or outside of the bend depending on the direction of propagation. The interaction between the presence of wall lining and curvature is investigated in particular; for instance, in a representative case it is found that the curvature causes the first few acoustic modes to be more heavily damped by the duct boundary than would be expected for a straight duct. Analytical progress can be made in the limit of very high mode order, in which case well-known 'whispering gallery' modes, localised close to the wall, can be identified.

Graphone model interpolation and Arabic pronunciation generation

This paper extends n-gram graphone model pronunciation generation to use a mixture of such models. This technique is useful when pronunciation data is for a specific variant (or set of variants) of a language, such as for a dialect, and only a small amount of pronunciation dictionary training data for that specific variant is available. The performance of the interpolated n-gram graphone model is evaluated on Arabic phonetic pronunciation generation for words that can't be handled by the Buckwalter Morphological Analyser. The pronunciations produced are also used to train an Arabic broadcast audio speech recognition system. In both cases the interpolated graphone model leads to improved performance. Copyright © 2011 ISCA.