Intelligibility of Vowels Altered in Duration and Frequency

Abstract

This study was conducted to examine the differential effects upon vowel intelligibility of various degrees of time compression (TC) and frequency division (FD), with and without time restoration. A male speaker and a female speaker were used. For a given percentage of distortion, FD degrades vowel intelligibility more severely than TC. Restoring time to normal for FD speech produced by slow playing via the Fairbanks speech compressor does not enhance intelligibility. Vowel confusions under TC are related to duration; those for FD conditions appear to be closely related to the perception of Vowel Formant 2 and to a lesser degree, Vowel Formant 1. Patterns of male and female vowel confusions are generally much alike for all conditions and types of distortion. Results tentatively indicate superior female vowel intelligibility under all conditions of distortion, the advantage being largest for FD, somewhat less for TC. These results suggest that over a limited range of frequency division (up to 40%), vowel phonemic quality is relatively unaffected by proportionate shifting of fundamental frequency and formant structure, indicating that a "relativevowel" hypothesis of vowel phonemic quality may hold for limited shifts in the frequency of vowel spectra. I. PROBLEM HE purpose of this study was to examine the effects of time compression (TC) and frequency division (FD) with (FD-TD) and without (FD-TR) time distortion upon vowel intelligibility. Much previous work with FD (Ochai, Saito, and Sakai, x Klumpp and Webster, •' Tiffany andBennett 3) has been limited by the fact that FD provided by slow-playing speech distorts time relationships. Use of the speech compression device built and described by Fairbanks, Everitt, and Jaeger 4 permits the restoration of FD speech to normal time (FD-TR) as was done by

Investigation of perceptual constancy in the temporal-envelope domain

“…The alternative, more plausible account of this discrepancy is that previous studies have compared categorization performance for reference and transposed signals under conditions where the former were always discriminable ͑e.g., Fairbanks and Kodman, 1957;Daniloff et al, 1968;Fu et al, 2001͒. It may then well be that, though degraded, performance for the transposed signal did not show a measurable drop due to a ceiling effect.…”

“…Indeed, identification of temporally transposed linguistic signals remains unaffected by transposition up to a compression/expansion index ͑CE͒ of 50% ͑e.g. Fairbanks and Kodman, 1957;Daniloff et al, 1968;Vaughan and Letowski, 1997;Gordon-Salant and Fitzgibbons, 2001;Versfeld and Dreschler, 2002͒. Some studies on categorical perception of phonemes also seem to provide evidence for the existence of some form of temporal normalization ͑Summer-field, 1981; Miller and Volaitis, 1989͒.…”

Proportional frequency compression of speech for listeners with sensorineural hearing loss

“…Daniloff et al ͑1968͒ found that vowel recognition for normal-hearing listeners remained at very high levels as long as the compression factor was no more severe than 0.7. They also found that the recognition of materials produced by female talkers was somewhat more resistant to frequency compression than that of male talkers.…”

Recognition of time-distorted sentences by normal-hearing and cochlear-implant listeners

“…When the words were shortened to 10% of their original duration, intelligibility approached 50% correct. Daniloff et al ͑1968͒ examined the effects of various degrees of time compression on vowel recognition. They found that time compression did not affect intelligibility appreciably until the phonemes were shortened to 30% of their original duration.…”