Appeal No. 2006-1690
Application No. 10/154,185
(ERLE of non-adaptive filter) merely indicate how well filters 21 and 22 perform in
canceling echo signals [brief, page 4].
Appellants further argue that Park actually teaches the opposite of
detecting double-talk in col. 5, line 62 where Park states that "it is not necessary
to detect double-talk" [id.] The examiner responds that Park's statement that "it
is not necessary to detect double-talk" in col. 5, line 62, when read in context,
"means the 'dual-h' echo canceller 20 does not diverge in the presence of
double-talk…" [answer, page 15].
Appellants also argue that Park in col. 5, lines 59-67 only discloses that
ERLE of the fixed filter 22 becomes higher during double-talk. According to
appellants, Park teaches nothing about when weight replacement is suspended
(i.e., avoiding step 891 in Fig. 8C in either steps 885 or 887) before portions of
the first signal are canceled by the non-adaptive filter [brief, page 5]. The
examiner responds that Park expressly indicates that when the ERLE of the fixed
filter (ERLE2) becomes higher than the adaptive filter's ERLE (ERLE1) during
double-talk, the fixed filter's coefficients are not updated with the adaptive filter's
coefficients, thus confirming operation of step 887 in Fig. 8C [answer, page 16].
The examiner notes that Park in Fig. 5 depicts the effects of double-talk on
ERLE1 and ERLE2 [answer, page 12]. Specifically, the examiner notes that Fig.
5 shows that the ERLE of the adaptive filter converges alongside the fixed filter's
ERLE during frames 0-20, but at frame 56, the adaptive filter's ERLE swings
below that of the fixed filter that remains relatively constant. The examiner
concludes that the only explanation for this divergence is that the adaptive filter
5
Page: Previous 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Next
Last modified: November 3, 2007