Stress detection: a sonic approach

We propose a transformation of digitalized spoken signal, that retains the prosodic characters of speech, and is robust under perturbations. Most of the existing signal processing for prosodic study is based on the assumption that the signal be continuous in time. Our transformation changes the signal to one completely discontinuous, dropping any semantic content except those related to rhythm and intonation. We put forward a characterization of stresses occurring after implosive consonants in Italian and in Brazilian Portuguese. The perceptual synchronization that characterizes it, is probably a more general phenomenon regarding "stresses" in other contexts. The transform is numerical, and "sonic", i.e. to be judged by trained ears, in the sense of native speakers of the language, for its robustness. The findings are on the objective transformed signal. -