Some effects of speaking rate on spectral and temporal characteristics of American English vowels Free

The effects of speaking rate on spectral and temporal characteristics of American English vowels were studied. Five women and five men, native American English speakers without strong regional dialects, produced the carrier phrase sentence “I said hVd again” at slow, conversational, and fast rates. Twelve vowels, /i, ɪ e^y, ɛ, æ ɜ, ʌ ɑ, ɔ, o^w, ʊ u/ were studied in /hVd/ context. Acoustic analyses were performed using the SPIRE speech analysis programs on a LISP machine. Recordings were digitized at a 16‐kHz sampling rate and low‐pass filtered. Spectral measurements included fundamental frequency and the first four formants. Temporal measurements made were vowel duration, syllable duration, and closure duration. Durations, duration differences, and rate related changes were quite systematic within and across vowel types. Results indicated that the durations of long vowels and diphthongs were compressed more than that of short vowels as speaking rate increased. Spectral differences were relatively small across speaking rates. The utility of duration as well as spectral information for vowel classification will be discussed. [Work supported by NIH and UTD research grants, and performed in part at the Research Lab. of Electronics, MIT.]