word corpus 0.0027964030000000003
word lexicon 0.0026464360000000003
oov word 0.002566177
english word 0.0025647070000000003
first word 0.002531607
ing word 0.002487685
word detection 0.002479485
word units 0.002465187
single word 0.002463929
word vocabulary 0.002443374
word segmentations 0.002418157
complete word 0.002378606
whole word 0.002377687
all word 0.0023770980000000002
word mod 0.002360584
generic word 0.002360336
model training 0.002358896
plete word 0.002357946
word anjani 0.002339589
cabulary word 0.002338063
training data 0.002205386
language model 0.002171989
segmentation model 0.002164802
hybrid model 0.002037078
other words 0.001980329
detection model 0.0019287949999999999
model parameters 0.001893667
data set 0.00188657
current model 0.001843743
probabilistic model 0.001825537
oov words 0.0018001669999999999
english words 0.001798697
new words 0.0017927029999999999
abilistic model 0.001787061
lectures data 0.001688578
vocabulary words 0.001677364
news data 0.0016765019999999999
complete data 0.001674406
development data 0.001653302
oovcorp data 0.001635126
data likeli 0.001633732
sampling words 0.001620869
likely words 0.001589441
known words 0.001584921
foreign words 0.001581463
unique words 0.001579924
eign words 0.00157267
pected words 0.00157267
model 0.00154336
language models 0.001543233
hybrid models 0.001408322
phone models 0.0013797319999999998
words 0.00132804
training set 0.001312256
acoustic models 0.001302844
training text 0.001240287
training learning 0.001232771
lvcsr models 0.001226163
phone context 0.0012033389999999999
test set 0.001200077
training oovs 0.0011996939999999998
possible corpus 0.001168368
flat models 0.001158575
context features 0.001131932
same lexicon 0.0011276839999999999
text corpus 0.0011271039999999999
lvcsr training 0.0011270949999999998
example corpus 0.001124907
discriminative training 0.001123167
hybrid language 0.001122347
training input 0.001115465
lexicon baseline 0.001099842
adaptive training 0.001099016
different lexicon 0.001099006
test oovs 0.001087515
detector training 0.00107193
detection performance 0.001070972
random segmentation 0.001065187
previous segmentation 0.001063442
first segmentation 0.0010589990000000001
hybrid lexicon 0.001046104
extended context 0.001045219
example segmentation 0.0010439960000000002
baseline hybrid 0.001041174
other unit 0.001037324
context fea 0.001021094
new lexicon 0.001017049
labeled corpus 0.001011458
left context 0.001002841
test sets 9.968450000000001E-4
standard detection 9.86925E-4
segmented corpus 9.86429E-4
context clues 9.82744E-4
overlapping context 9.82744E-4
morphological segmentation 9.81536E-4
baseline system 9.78744E-4
other segmentations 9.76396E-4
detection approach 9.75704E-4
corpus priors 9.74089E-4
baseline results 9.739449999999999E-4
