language model 0.00488844
acoustic model 0.003924994
bayesian model 0.003883698
model parameters 0.003778562
sequential model 0.003758437
mixture model 0.003742915
triphone model 0.003735427
model the 0.003721404
model pmm 0.003713815
markov model 0.003688886
grapheme model 0.003685294
our model 0.003684436
model unit 0.003680056
model retraining 0.003678889
model base 0.00367558
tic model 0.003670079
model gener 0.003655302
tive model 0.003649204
model implementation 0.003648588
model refining 0.003646136
model outper 0.003642976
quential model 0.003642419
model conventionally 0.003642419
model 0.00341801
word pronunciation 0.0027907
word error 0.002645067
ing word 0.00252067
word pronunciations 0.002499343
word tokens 0.002420302
corresponding word 0.002360874
arbitrary word 0.002354591
word transcription 0.002351148
word transcriptions 0.002350254
word burma 0.00234571
word pronuncia 0.002340239
word pronunci 0.002334598
word fox 0.002331461
word lake 0.002331461
language models 0.002320064
other language 0.002004972
new language 0.0019396629999999999
speech data 0.001921406
trigram language 0.001747338
speech corpus 0.0017154980000000002
language barrier 0.001694604
phone speech 0.001599195
other words 0.001496411
language 0.00147043
speech recognition 0.001421224
stochastic speech 0.001416495
speech feature 0.00140428
large speech 0.001403198
speech features 0.0013726180000000001
speech utterance 0.0013651000000000002
acoustic models 0.001356618
speech rec 0.0013443790000000001
monophone speech 0.001343507
speech recognizers 0.0013400200000000001
automatic speech 0.001338224
speech segments 0.001338176
speech recognizer 0.0013344720000000002
annotated speech 0.001333083
data set 0.001326255
speech signal 0.0013200400000000002
speech segment 0.0013152910000000001
speech recogni 0.001298729
pronunciation lexicon 0.001283341
following distribution 0.0012700839999999999
frequent words 0.001228809
dirichlet distribution 0.0012081169999999999
recognition models 0.001200288
stochastic models 0.0011955590000000001
mixture models 0.001174539
phonetic set 0.00116829
text data 0.001168197
triphone models 0.0011670510000000001
cal distribution 0.001164587
training set 0.001132321
mapping distribution 0.001107929
entire data 0.001105462
tic models 0.001101703
posterior distribution 0.001089896
many pronunciation 0.001083292
uniform distribution 0.001082744
pronunciations pronunciation 0.0010763230000000001
single pronunciation 0.00107324
speech 0.00107057
phonetic units 0.0010618799999999999
emission distribution 0.0010579419999999999
pronunciation rules 0.001050291
training results 0.001037075
relative probability 0.001026439
segmentation method 0.001023006
pronunciation list 0.001014823
phonetic boundaries 0.001010603
pronunciation mixture 0.0010087450000000001
new phone 9.97858E-4
recognition system 9.95518E-4
pronunciation knowledge 9.88524E-4
phone boundary 9.84761E-4
