Key points are not available for this paper at this time.
This paper dacribes the results of some simple vowel-discrimination experiments based on isolated trames of data.That data was produced from short segments taken from the TIMIT database, and took the form of the new, linear frequency-scale, version of the Reduced Auditory Representation (BAR) l.The recognition was performed using standard linear discriminant techniques, albeit with arather lower dimensionality than normal heuuse of the reduced number of classes which were to be identi ed.
BEET et al. (Tue,) studied this question.