The perception of speech rate from the sound-silence patterns of monologues

Abstract

Two experiments asked whether listeners can judge word rate from a speech signal that has been degraded in various ways. In the first, the rates of spontaneous speech were increased by 42% and further transformed to produce tone-silence sequences. The tonesilence sequences were presented to listeners who judged the rate of each sequence. Results clearly indicated that listeners could differentiate the rates of the tone-silence sequences, suggesting that minimal nonlinguistic information may be sufficient to make grossly accurate estimates of speech rates. In the second study, listeners were presented with speech sequences involving three naturally produced rates (slow, moderate, and fast) in three conditions (clear, frequency-inverted, and tone-silence) such that different listeners participated in the three conditions, but heard all rates in each condition. Listeners in the clear and frequency-inverted conditions distinguished all three rates, but those in the tone-silence condition differentiated only the slow and moderate rates. Contrary to expectation, the gender and extroversion scores of the listeners did not affect their judgments.