We consider the problem of detecting stop consonants in continuously spoken speech. We pose the problem as one of finding the optimal filter (linear or non-linear) that operates on a particular appropriately chosen representation. We discuss the performance of several variants of a canonical stop detector and consider its implications for human and machine speech recognition.

Consonants and vowels differ in the extent to which they are perceived categorically. We use a Bayesian model of speech perception to explore factors that might cause this difference. Simulations show that perception of vowels, fricatives, and stop consonants can all be captured under a single model in which listeners use their knowledge of phonetic categories to infer the sound that a speaker ...

An incomplete stop consonant is characterized either by an indistinguishable closure or a missing burst. If an incomplete stop happens due to a stop following another stop [stop-stop interaction (SSI)], its acoustics typically resemble that of a complete stop-one closure followed by a single burst. As a consequence, stop detectors would fail to distinguish an SSI from a complete stop. Analysis ...

The possibility of using a circuit to amplify selectively the burst of a stop consonant is investigated. It is shown that such a circuit used in the speech channel of an amplifying system can improve the discrimination between stop consonants. Such a system could be of value in assisting the hard of hearing.

A set of auditorily-formulated features for PLACE discrimination in stop consonants, uncovered in extensive experiments with natural and edited sounds, are now being modeled using fuzzy logic and being applied to large databases of monosyllabic and spelled letters speech sounds, in various languages, in full-band and low-pass conditions. The rationale is that any valid model of human communicat...