| 摘要 |
PURPOSE: To realize thinning display without damaging features of a voice waveform by extracting feature points in each fixed section of a digital time series signal of an inputted voice and displaying a sequence of feature points on a display as a voice waveform. CONSTITUTION: A voice time series signal having, for example, 20kHz sampling frequency is inputted to a voice waveform input part 11. A section division part 12 divides the voice waveform signal sent from the input part 11 by 1msec, namely, 20 samples. A maximum value extraction part 12b and a minimum value extraction part 12c extract a maximum value and a minimum value from voice waveform data of 20 samples respectively. A comparison part 12d receives the maximum value Dmax extracted by the extraction part 12b and the minimum value Dmin extracted by the extraction part 12c to compare their absolute values with each other and outputs data having the larger absolute value. This operation is repeated for each section, and voice data outputted from the comparison part 12d is subjected to signal processing by a voice waveform display part 13 and is displayed on a display device 14, thus realizing thinning display of the voice waveform. |
