摘要 |
PROBLEM TO BE SOLVED: To surely identify a consonant section in an audio signal through simple signal processing of light processing load.SOLUTION: A sample sequence representing a waveform of an audio signal is divided into blocks as many as the predetermined number of samples, energy of all frequency components in the blocks and energy of a frequency component belonging to a predetermined frequency band in the blocks is calculated, respectively, for each block and ratio of both is calculated. In addition, the number of zero crosses per unit time in each block is counted. On the basis of the energy ratio and a variation mode of the number of zero crosses between neighboring blocks, it is then determined whether or not the blocks are included in a consonant section for each block. |