| 摘要 |
The invention relates to a method for detecting a transient in a discrete-time audiosignal (x(k)) which is carried out entirely in the time domain. Said method comprises a step in which the discrete-time audiosignal is segmented so as generate consecutive segments of identical length with unfiltered discrete-time audiosignals (xs(T), xs(T-1), xs(T-2), ...), after which the discrete-time audiosignal (xs(T)) in a current segment is filtered. Thereafter there are two options: either the energy (Ef(T)) of the filtered discrete-time audiosignal (Ys(T)) in the current segment can be compared with the energy (Ef(T-1)) of the filtered discrete-time audiosignal (Ys(T-1)) in a preceding segment, or a current relationship can be generated between the energy (Ef(T)) of the filtered discrete-time audiosignal (Ys(T)) in the current segment and the energy (Eu(T)) of the unfiltered discrete-time audiosignal (Xs(T)) in the current segment and said current relationship compared with a corresponding preceding relationship. On the basis of the one comparison and/or the other comparison it is determined whether a transient is present in the discrete-time audiosignal.
|
| 申请人 |
FRAUNHOFER-GESELLSCHAFT ZUR FOERDERUNG DER ANGEWANDTEN FORSCHUNG E.V., 80636 MUENCHEN, DE |
发明人 |
HILPERT, JOHANNES, DIPL.-ING., 90763 FUERTH, DE;HERRE, JUERGEN, DIPL.-ING., 91054 BUCKENHOF, DE;GRILL, BERNHARD, DIPL.-ING., 91207 LAUF, DE;BUCHTA, RAINER, DIPL.-ING., 91074 HERZOGENAURACH, DE;BRANDENBURG, KARL-HEINZ, DR.-ING., 91054 ERLANGEN, DE;GERHAEUSER, HEINZ, DR.-ING., 91344 WAISCHENFELD, DE |
