DYNAMIC TIME WARPING ARRANGEMENT

摘要

DYNAMIC TIME WARPING ARRANGEMENT A first speech pattern and a second speech pattern are aligned so that a global similarity signal assumes an extremum. For a current first speech pattern time frame, second speech pattern time frames are selected. A predicted global similarity signal is generated between the current first speech pattern time frame and each of the selected second speech pattern time frames. Each first speech pattern time frame is paired with a second speech pattern time frame responsive to the predicted global similarity signals.