Abstract
A method of synchronizing speech waveform playback and text display is disclosed. The synchronization can be performed in the syllable, character or word level. The method includes the following approaches: getting input text, which includes multiple syllables, characters and words; getting the reference feature vector sequence according to the input text by concatenating multiple reference feature vector sequences which are from a database of feature vector sequences of all linguistic units (like syllables, characters and words) for the target language or languages; getting a speech waveform; extracting the feature vector sequence from the speech waveform; searching for the syllable boundaries by aligning the extracting feature vector sequence and the reference feature vector sequence, where the alignment is performed by using the dynamic time warping (DTW) technique.
Translated title of the contribution | METHOD OF SYNCHRONIZING SPEECH WAVEFORM PLAYBACK AND TEXT DISPLAY |
---|---|
Original language | Chinese (Traditional) |
Patent number | I269191 |
IPC | G06F-017/20(2006.01) |
State | Published - 21 12 2006 |
Bibliographical note
公開公告號: I269191Announcement ID: I269191