No abstract

A sentence verification task (SVT) was used to study the effects of sentence predictability on comprehension of natural speech and synthetic speech that was controlled for intelligibility. Sentences generated using synthetic speech were matched on intelligibility with natural speech using results obtained from a separate sentence transcription task. In the main experiment, the sentence verification task included both true and false sentences that varied in predictability. Results showed differences in verification speed between natural and synthetic sentences, despite the fact that these materials were equated for intelligibility. This finding suggests that the differences in perception and comprehension between natural and synthetic speech go beyond segmental intelligibility as measured by transcription accuracy. The observed differences in response times appear to be related to the cognitive processes involved in understanding and verifying the truth value of short sentences. Reliable effects of predictability on error rates and response latencies were also observed. High-predictability sentences displayed lower error rates and faster response times than low-predictability sentences. However, predictability did not have differential effects on the processing of synthetic speech as expected. The results demonstrate the need to develop new measures of sentence comprehension that can be used to study speech communication at processing levels above and beyond those indexed through transcription tasks, or forced-choice intelligibility tests such as the Modified Rhyme Test (MRT) or the Diagnostic Rhyme Test (DRT).Over the past six years, numerous studies on the perception of synthetic speech have been conducted in our laboratory at Indiana University (Pisoni, 1982;Pisoni, Nusbaum & Greene, 1985). The bulk of these studies have focused on measures of segmental intelligibility, such as identification of isolated words and recognition of words in sentences (e.g., Egan, 1948;House, Williams, Hecker & Kryter, 1965;Nye & Gaitenby, 1973). Results from these studies of phoneme and word perception have shown that synthetic speech is consistently less intelligible than natural speech (Greene, Logan & Pisoni, 1986). This finding was observed for a variety of synthesis systems ranging from very low-quality, low-intelligibility products, such as the ECHO and Votrax Type N Talk, to extremely natural sounding speech with very high intelligibility such as DECtalk and the Prose 2000.Most perceptual studies dealing with segmental intelligibility have not addressed the issue of comprehension processes involved in understanding the linguistic content of the message. In tests of segmental intelligibility, such as the ones we have carried out, subjects are not NIH Public Access Author ManuscriptComput Speech Lang. Author manuscript; available in PMC 2012 December 05.Published in final edited form as:Comput Speech Lang. 1987 ; 2(3-4): 303-320. NIH-PA Author ManuscriptNIH-PA Author Manuscript NIH-PA Author Manuscript required to e...

A verification task was used to study sentence comprehension using two natural voices and five different synthetic voices generated by automatic text-to-speech conversion. Subjects listened to true and false, three- and six-word sentence produced by one of these voices. Sentence verification accuracy and speed, and sentence transcription accuracy were measured. A significant effect of voice type was obtained for true and false sentences for all three measures. In addition, for false sentences, subjects were less accurate in transcribing six-word sentences and they were slower in verifying these sentences. Furthermore, there were significant interactions of voice type with sentence length for all three dependent measures. Verification speed revealed a clustering of the seven voices into three basic categories corresponding to: (1) natural speech, (2) high-quality synthetic speech, and (3) moderate- to low-quality synthetic speech. Results from a second sentence verification task used to investigate effects of sentence predictability will also be discussed. [Work supported by AFOSR and NIH.]

No abstract