Appeal No. 1996-2198
Application 08/077,506
G. The merits of the § 103 rejection of
claims 1 and 3-5 over Kuroda in view of Baker
The examiner reads the first three elements of claim 1
onto Kuroda's HMM speech recognition system as follows:
Claim Kuroda
"feature analyzer" ........ feature extraction circuit 4
"table" ................... parameter table 11
"converting means" ........ labeling circuit 5
We agree with appellants (Brief at 29) that the
examiner's reliance on Kuroda's parameter table as the claimed
table is incorrect. It appears to us, and appellants do not
deny, that this limitation corresponds instead to Kuroda's
label prototype dictionary 6, from which labeling block 5
selects the prototype that is closest to the feature extracted
by feature extractor 4 (col. 3, lines 27-50).
Next, the examiner reads the claimed first forming means
for forming the state transition probability (A) and the
probability distribution (B) of occurrence of the feature
codes onto Kuroda's training block 8 and adaptation block 9.
Appellants concede (Brief at 31 & n.17) that Kuroda's training
-16-
Page: Previous 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Next
Last modified: November 3, 2007