摘要 |
Keyword based audio data comparison. A method comprises extracting at least one feature from audio content of an audio data set to form feature arrays. The feature arrays may comprise a pitch array and a rhythm array. A set of keywords is formed for each feature array by performing the following for each feature array. Keywords are formed from overlapping sequences of values in the feature array, wherein each successive keyword comprises at least one value of a previous keyword and further comprises at least one value from the feature array that is not in the previous keyword. Another aspect audio is keyword-based audio data searching. Search keywords are formed from audio content. An index comprising sets of keywords formed from audio content of respective ones of audio data sets is searched using the search keywords formed from audio content of the specified audio data to identify audio data having similar audio content.
|