摘要 |
PURPOSE:To vocalize any word at a high speed with superior sound quality by thinning out only frames which are stable in pitch period, etc., in a vowel section. CONSTITUTION:This speech synthesizer edits and synthesizes plural speech information parameters including spectrum information, analyzed every frames of a certain time length, in the unit of a frame and is equipped with a speech memory 2 which stores speech data (a) from a speech file 1, a predicted gain calculator 3 which calculates the predicted gain of spectrum information (b), and a decision unit 4 which compares the predicted gain and its variation quantity with threshold values. Further, this synthesizer is equipped with a buffer control circuit which controls a buffer memory 6 and a buffer memory 7 with the output of the decision unit 4, a synthesizing filter 8 which puts the outputs of the buffer memories 6 and 7 together, and a frame control circuit 10 which thins out frames by controlling the speech memory 2. |