Speech science sings a new tune

Two physicists at King’s College London have discovered a precise way to model the natural resonance frequencies, called formants, of the human vocal tract. Barbara Forbes (also of the startup Phonologica) and Roy Pike applied the methods of quantum mechanics, in the form of a Klein–Gordon equation, to the classical acoustics of a simple organ pipe, which speech researchers often study to gain basic insights into sound production in humans. A perturbation analysis showed that adding curvatures—dents or bumps—at optimal positions in a straight pipe can shift its natural resonance frequencies up or down, largely independently of each other. The analysis substantially advances Lord Rayleigh’s long-standing 1878 result, which looked at a pipe’s perturbed cross section but did not account for the resulting wave dispersion. The researchers are now using their theory to program a machine to recognize natural phonetic sounds. They also say the theory has applications in engineering, musical acoustics, and physiological speech production. (B. J. Forbes, E. R. Pike, Phys. Rev. Lett. 93 , 054301, 2004 http://dx.doi.org/10.1103/PhysRevLett.93.054301 .)