ERIC Number: ED021851
Record Type: RIE
Publication Date: 1968-Jul
Reference Count: 0
The Intelligibility of Time-Compressed Speech. Final Report.
Carroll, John B.; Cramer, H. Leslie
Time-compressed speech is now being used to present recorded lectures to groups at word rates up to two and one-half times that at which they were originally spoken. This process is particularly helpful to the blind. This study investigated the intelligibility of speech processed with seven different discard intervals and at seven rates from two to five times the original. Optimal parameters for processing speech at each of the rates are reported along with a comparison of intelligibility rendered by three different modes of presentation. Headphone presentation was found to be about 22% more intelligible on the average than presentation by loudspeaker. The commercial equipment available for processing time-compressed speech varies the sampling interval but maintains a constant discard interval. The findings indicate that the sampling interval should be held constant for rates two to five times normal. A sampling interval of 15 milliseconds was found to be optimal at all rates for a low-pitched man's voice. A higher pitched woman's voice would require an even shorter sampling interval. Processing and presenting time-compressed speech using the optimal parameters reported here should promote greater comprehension as well as intelligibility at all rates, intelligibility being a necessary prerequisite for comprehension. (Author)
Publication Type: N/A
Education Level: N/A
Sponsor: Office of Education (DHEW), Washington, DC. Bureau of Research.
Authoring Institution: Harvard Univ., Cambridge, MA. Lab. for Research in Instruction.