Temporally segmented speech

Abstract

Temporally segmented speech is continuous speech broken up by the insertion of silent intervals. The durations of the resulting speech intervals and silent intervals can be varied independently. When silent intervals are held constant at 200 msec, and speech interval duration is varied, intelligibility falls from about 90% to about 10% as speech interval duration is reduced from 200 to 30 msec. When speech interval duration is held constant at 63 msec, and silent interval duration is varied, intelligibility recovers from its asymptotic value of about 50% with long silent intervals, to 100% as the silent intervals are shortened from about 120 msec to about 60 msec. Implications for short-term acoustic storage are discussed.