Perceived speech rhythm tracks the loudness-time function of the signal

Evidence

The temporal intervals between the neighbouring events of the reproduced rhythm have been measured for German, English, French and Japanese. A histogram of all reproduced time intervals based on more than 10000 data points is indicated in Fig. 16.46 with a bin width of 20 ms. All the intervals are concentrated between 100 and 1000 ms. Almost 90% of the data points are found between 200 and 600 ms, and the most frequent temporal interval has a length of about 300 ms. This means that in running speech, about 2 to 5 events are perceived per second. These data correspond nicely to the maximum of the fluctuation strength measured as a function of the modulation rate. The sequence of processing is evident if the loudness-time function is considered to be the initial basic information. The maxima of this function correlate with the perceived rhythm. The segmentation based on this strategy leads either to syllables or half syllables if the minima in the loudness-time function are used as well. — Professor Dr.-Ing. Hugo Fastl, Professor Dr.-Ing. Eberhard, Psychoacoustics- Facts and Models, 2035

Categorization

• DEFINE: Explains how the speech rhythm a listener perceives is derived from segmentation of the loudness-time envelope, an explicit processing model.