Second-order recurrent networks that recognize simple finite state languages over {0,1}* are induced from positive and negative examples. Using the complete gradient of the recurrent network and sufficient training examples to constrain the definition of the language to be induced, solutions are obtained that correctly recognize strings of arbitrary length.
Spectrographic data are presented which suggest that it may be possible to estimate the frequency of the fundamental resonance of the cavity behind the mouth opening, the ’’front cavity resonance,’’ from information in the speech signal. It is shown that place of articulation information in the steady states, transitions, and bursts of F2 (or sometimes F3) can be reinterpreted to be information from the front cavity resonance. Furthermore, a number of synthesis results that have appeared anomalous when described in terms of numbered formants seem to find a coherent explanation in terms of the front cavity resonance. Implications for theories of speech perception include the possibility that an estimate of front cavity resonance frequency may serve for continuous articulatory reference.
Subject Classification: 70.20, 70.30.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.