| 摘要 |
PURPOSE:To stably and naturally change the speech speed of inputted audio signals from 'slow' to 'fast'. CONSTITUTION:While expanding the voiced sound segment of inputted audio signals in accordance with a certain rule, a expansion magnification function g(t,w), that is a function of a segment length w of the voiced sound segment and time t at which the segment appears, is used to change the value of an expansion magnification. For example, the magnification function g(t, w), which is shown in Figure 3 and is a function of w and t, is applied to a short voiced sound, which appears at a starting point of a phrase or appears <=450ms after a pitch change exceeds a certain value. If the length of a voiced sound segment is less than 150ms as w1 shown in the figure, the magnification at the time of the end of a voiced sound segment is applied during the segment. On the other hand, if the voiced sound segment exceeds 150ms as w2, the segment is divided into 150ms units and the magnification which corresponds to the time of the respective ending point is applied. |
