C88-1082 |
also carried out using bottom-up
|
phoneme recognition
|
results \ -LSB- 12 \ -RSB- .
|
H89-2047 |
results for speaker-independent
|
phoneme recognition
|
. We performed experiments on
|
N09-1021 |
due to the considerably improved
|
phoneme recognition
|
afforded by longer recognition
|
C88-1082 |
parameters are extracted and bottom-up
|
phoneme recognition
|
is carried out . The phrase hypotheses
|
C88-1082 |
feature extraction part and a
|
phoneme recognition
|
part \ -LSB- 10,11 \ -RSB- .
|
P08-2042 |
manual transcription . Over 85 %
|
phoneme recognition
|
accuracy is demonstrated for
|
C88-1082 |
from errors and rejections in
|
phoneme recognition
|
and greatly increases the accuracy
|
P05-1064 |
P-PRLM . Figure 1 . L monolingual
|
phoneme recognition
|
front-ends are used in parallel
|
C88-1082 |
. In such methods , however ,
|
phoneme recognition
|
errors and rejections result
|
H94-1061 |
would appear that although higher
|
phoneme recognition
|
rates are achieved for French
|
P08-5004 |
parsing , entity extraction , and
|
phoneme recognition
|
. Our algorithmic framework will
|
C88-1082 |
and phrase recognition based on
|
phoneme recognition
|
, the parser extracts the sentence
|
H89-2033 |
was used very effectively in a
|
phoneme recognition
|
system ( McDermott , 1989 ) .
|
N09-1021 |
from the low accuracy typical of
|
phoneme recognition
|
. We consider two methods for
|
H89-2047 |
that this result also holds for
|
phoneme recognition
|
, when phoneme segmentation boundaries
|
N09-1021 |
using standard word-based LVCSR ,
|
phoneme recognition
|
, and LVCSR using phoneme multigrams
|
J88-2015 |
+65 dB , measured 64 % correct
|
phoneme recognition
|
. It was estimated that an effective
|
H92-1100 |
the dynamical system model in
|
phoneme recognition
|
( as opposed to classification
|
H89-2047 |
the Stochastic Segment Model for
|
Phoneme Recognition
|
</title> V Digalakis t M Ostendofft
|
C88-1082 |
phrase hypotheses for top-down
|
phoneme recognition
|
are pre-selected by the { 1 }
|