# include common settings for fullLP systems.
. conf/common.fullLP || exit 1;

#speech corpora files location
train_data_dir=/export/babel/data/104-pashto/release-current/conversational/training
train_data_list=/export/babel/data/splits/Pashto_Babel104/train.FullLP.list
train_nj=32

#RADICAL DEV2H data files
dev2h_data_dir=/export/babel/data/104-pashto/release-current/conversational/dev
dev2h_data_list=/export/babel/data/splits/Pashto_Babel104/dev2hr.list
dev2h_data_cmudb=/export/babel/data/splits/Pashto_Babel104/uem/db-v7_dev+eval-utt.dat
dev2h_stm_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-dev/IARPA-babel104b-v0.4bY_conv-dev.stm
dev2h_ecf_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-eval.ecf.xml
dev2h_rttm_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-dev/IARPA-babel104b-v0.4bY_conv-dev.mitllfa3.rttm
dev2h_kwlist_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-dev/IARPA-babel104b-v0.4bY_conv-dev.kwlist2.xml
dev2h_subset_ecf=true
dev2h_nj=18

#Official DEV data files
dev10h_data_dir=/export/babel/data/104-pashto/release-current/conversational/dev
dev10h_data_list=/export/babel/data/splits/Pashto_Babel104/dev.list
dev10h_data_cmudb=/export/babel/data/splits/Pashto_Babel104/uem/db-v7_dev+eval-utt.dat
dev10h_stm_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-dev/IARPA-babel104b-v0.4bY_conv-dev.stm
dev10h_ecf_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-eval.ecf.xml
dev10h_rttm_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-dev/IARPA-babel104b-v0.4bY_conv-dev.mitllfa3.rttm
dev10h_kwlist_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-dev/IARPA-babel104b-v0.4bY_conv-dev.kwlist2.xml
dev10h_nj=32


#Official EVAL period evaluation data files
eval_data_dir=/export/babel/data/104-pashto/release-current/conversational/eval/
eval_data_list=/export/babel/data/splits/Pashto_Babel104/eval.babel104b-v0.4bY.list
eval_data_cmudb=/export/babel/data/splits/Pashto_Babel104/uem/db-v7_dev+eval-utt.dat
eval_ecf_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-eval.ecf.xml
eval_kwlist_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-eval.kwlist2.xml
eval_nj=64

#Official (POST-)EVAL evaluation data portion
#evalpart1_data_dir=/export/babel/data/104-pashto/release-current/conversational/eval
#evalpart1_data_list=
#evalpart1_data_cmudb=/export/babel/data/splits/Pashto_Babel104/uem/db-v7_dev+eval-utt.dat
#evalpart1_stm_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-evalpart1/IARPA-babel104b-v0.4bY_conv-evalpart1.stm
#evalpart1_ecf_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-evalpart1.ecf.xml
#evalpart1_rttm_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-evalpart1/IARPA-babel104b-v0.4bY_conv-evalpart1.mitllfa3.rttm
#evalpart1_kwlist_file=/export/babel/data/scoring/IndusDB/IARPA-babel104b-v0.4bY_conv-evalpart1/IARPA-babel104b-v0.4bY_conv-evalpart1.annot.kwlist2.xml
#evalpart1_nj=32

# Acoustic model parameters
numLeavesTri1=1000
numGaussTri1=10000
numLeavesTri2=1000
numGaussTri2=20000
numLeavesTri3=6000
numGaussTri3=75000
numLeavesMLLT=6000
numGaussMLLT=75000
numLeavesSAT=6000
numGaussSAT=75000
numGaussUBM=800
numLeavesSGMM=10000
numGaussSGMM=80000

# Lexicon and Language Model parameters
oovSymbol="<unk>"
lexiconFlags="--romanized --oov <unk>"

# Scoring protocols (dummy GLM file to appease the scoring script)
#glmFile=./conf/glm
lexicon_file=/export/babel/data/104-pashto/release-current/conversational/reference_materials/lexicon.txt


#keyword search settings
duptime=0.5
case_insensitive=true