摘要 |
A plurality of pairs of segments to be weighted/added are selected non-linearly with respect to a time axis of audio data. A speed conversion is achieved by performing the weighting/addition on the selected pairs of segments. The non-linear selection is performed by (a) obtaining all possible pairs of segments constituting the audio data, (b) calculating a degree of similarity pertaining to each possible pair, (c) ranking the all possible pairs of segments according to the degrees of similarity, and (d) overlapping at least one of the all possible pairs of segments that holds the highest degree of similarity. |