lexicon word 0.0034851459999999997
word segmentation 0.003369806
new word 0.003184212
corresponding word 0.002999616
own word 0.002943348
word edge 0.002913103
word lattices 0.002889676
word lattice 0.0028854749999999998
lexicon words 0.002406486
language model 0.002354016
acoustic model 0.002252341
model perplexity 0.002118131
new character 0.002114782
new words 0.002105552
character error 0.00204348
oov words 0.00201871
model scores 0.002010563
model score 0.002005731
character accuracy 0.001987887
character recognition 0.001937939
character strings 0.001930991
average character 0.001889663
asr character 0.001888848
output words 0.00186894
lexical words 0.001856883
character posterior 0.001845275
improved character 0.0018283610000000001
character arc 0.001814991
character recog 0.001811813
character poste 0.001811813
character pps 0.001811813
character conversion 0.001811813
model 0.00173102
character 0.00157759
words 0.00156836
chinese language 0.001385345
new lexicon 0.001375318
corpus lexicon 0.0013539329999999999
other lexicon 0.001346548
lexicon adaptation 0.001276186
chinese case 0.001185984
lexicon bigram 0.001164656
previous lexicon 0.0011571160000000001
current lexicon 0.001127748
full lexicon 0.001092294
same approach 0.001091007
different point 0.00107376
chinese asr 0.0010736069999999999
criminative lexicon 0.0010724950000000001
line lexicon 0.0010724950000000001
new corpus 0.001052999
different statistics 0.0010392790000000002
same clustering 0.001018143
end time 0.001006189
ning time 0.001006189
ing set 9.87487E-4
speech recognition 9.78937E-4
adaptation corpus 9.53867E-4
posterior probability 9.44769E-4
standard way 9.34754E-4
mutual probability 9.21194E-4
such procedure 9.16679E-4
acoustic features 9.138150000000001E-4
ward probability 9.11404E-4
probability masses 9.11404E-4
adaptation approach 9.062899999999999E-4
automatic speech 8.97074E-4
statistical rules 8.89631E-4
new probabilities 8.82788E-4
speech decoding 8.803369999999999E-4
same history 8.587660000000001E-4
possible set 8.56655E-4
whole speech 8.53822E-4
clustering method 8.452710000000001E-4
training manual 8.43475E-4
lexicon 8.38126E-4
similar results 8.318399999999999E-4
training corpora 8.31212E-4
testing set 8.11223E-4
oov problem 8.087820000000001E-4
acoustic likelihood 8.05629E-4
adaptation problem 7.96492E-4
full set 7.93792E-4
different 7.92067E-4
such alignment 7.85345E-4
such arguments 7.798200000000001E-4
output units 7.7806E-4
asr error 7.771480000000001E-4
transcription characters 7.74631E-4
recognition accuracy 7.70646E-4
acoustic signals 7.68682E-4
several methods 7.675119999999999E-4
other hand 7.648900000000001E-4
programming method 7.63775E-4
chinese 7.62349E-4
error rate 7.60634E-4
text level 7.50344E-4
recognition results 7.4871E-4
oov rate 7.45094E-4
adaptation approaches 7.40401E-4
