摘要 |
PURPOSE: To emphasize the consonant part of a speech in parallel to a waveform expanding process for speech-rate conversion by performing the process for emphasizing the consonant part of the speech simultaneously by the speech-rate conversion process. CONSTITUTION: In the speech-rate conversion process 30 of the speech-rate conversion device, a consonant detecting means 33 which detects the consonant part from the time-series pattern of power values of input speech data and a consonant emphasizing means 34 which amplifies the value of the speech data at the consonant part in a detected frame are provided. The consonant detecting means 33 observes a time series of power values of speech data in the input frame as one pattern and matches it with a predetermined power pattern to detect the consonant part in the input frame. The consonant emphasizing means 34 partially emphasizes the value of the speech data in the frame according to the detection of the consonant detection and emphasizes the consonant part. Further, a power decision process 31 decides a frame for which a waveform expanding process 32 should be performed according to the power of the whole frame. |