| 摘要 |
PROBLEM TO BE SOLVED: To largely reduce expansion processing, and moreover, to easily change a speaking speed of the speech data received by an expansion side as a speech data separated based on a pitch period. SOLUTION: On a speech data compression side, a breakpoint detecting part 13 detects breakpoints on the basis of a pitch period from a speech waveform; a processing object fundamental wave acquisition part 13 takes out a compression processing object fundamental waveform from successive plural fundamental waveforms based on a certain rule defining a waveform between each breakpoint adopted as the fundamental waveform; and about the processing object fundamental waveform thus obtained, a compressed data generation means 14 executes encoding processing and creates a compressed data. And, a speech data expansion side receives compressed data for each processing object fundamental waveform; decodes the coded data existing in the compressed data by speech data decoding 22; and a reproduction speech creating part 23 reproduces a speech data to be reproduced, by overlapping the decoded fundamental waveforms in the time base direction. |
