Within-session stability of acoustic features of conversational and clear speech

Abstract

In several planned studies, talkers will perform a set of speech tasks several times in each recording session. For example, one study will have talkers perform a set of tasks four times: in quiet, in two levels of noise, and in reverberation. It is unknown, however, whether or how much the acoustic details of speech are affected by simple repetition of the speech task. Over the course of a recording session, talkers' speech might become less careful due to fatigue, boredom, familiarity with the speech materials, or some combination of these factors. Such repetition effects could offset the effects of speaking style instructions given at the beginning of a session. The present study assessed speech acoustic changes over four repetitions of a speech production task set performed under conversational or clear speech instructions. Sixteen talkers performed three speech production tasks (a passage, a list of sentences, and a picture description) four times in each condition. The two speaking styles were recorded in separate test sessions. Several acoustic features relevant to clear speech (e.g., vowel space and speaking rate) will be compared between the first and fourth repetition in each speaking style as well as between the styles.