Appeal No. 2006-1690
Application No. 10/154,185
provides updated coefficients to the fixed filter when double-talk is not detected,
but that the fixed filter is not updated when the presence of double-talk is
detected [id.].
The examiner further notes that comparing the power-based measure
ERLE1 with a threshold TE in block 855 in Fig. 8B corresponds to double-talk
detection because such a comparison blocks copying coefficients from the
adaptive filter to the fixed filter in steps 857 and 885 [answer, page 15].
Appellants respond that block 855 does not disclose detecting double-talk and
note that in Fig. 6 in Park, no double-talk is present, yet ERLE1 also drops below
threshold TE [reply brief, page 2].
We will sustain the examiner's anticipation rejection of claims 1, 8, 10-12,
14, 18, 19, 21, 23, and 27. We find that the echo canceler of Park detects
double-talk and suspends replacement of the non-adaptive filter weights before
portions of the first signal are cancelled by the non-adaptive filter as claimed.
When double-talk occurs in Park, ERLE1 is below ERLE2. See Park, Fig.
5 (noting the presence of double-talk from the fifty-sixth frame to the sixty-eighth
frame). Therefore, at step 887 in Fig. 8C, the coefficient from the fixed filter is
copied to the adaptive filter since step 895 is executed. Such a result necessarily
suspends replacement of the non-adaptive filter weights because step 891
(copying the adaptive filter's coefficient to the fixed filter) is not executed in this
condition (i.e., when double-talk occurs). See also Park, col. 6, line 66 - col. 7,
line 19 (noting that during double-talk, the fixed filter is used and maintains its
coefficient previously copied from the adaptive filter). Therefore, Park detects
6
Page: Previous 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Next
Last modified: November 3, 2007