Blame view
egs/vystadial_en/s5/run.sh
7.37 KB
8dcb6dfcb
first commit
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 |
#!/bin/bash
# Copyright Ondrej Platek Apache 2.0
renice 20 $$
# Load training parameters
. ./env_voip_en.sh
# Source optional config if exists
[ -f env_voip_en_CUSTOM.sh ] && . ./env_voip_en_CUSTOM.sh
. ./path.sh
# If you have cluster of machines running GridEngine you may want to
# change the train and decode commands in the file below
. ./cmd.sh
#######################################################################
# Preparing acoustic features, LMs and helper files #
#######################################################################
echo " Copy the configuration files to $EXP directory."
local/save_check.sh $EXP $WORK/* || exit 1;
local/download_en_data.sh $DATA_ROOT || exit 1;
local/data_split.sh --every_n $EVERY_N $DATA_ROOT $WORK/local "$LMs" "$TEST_SETS" || exit 1
local/create_LMs.sh $WORK/local $WORK/local/train/trans.txt \
$WORK/local/test/trans.txt $WORK/local/lm "$LMs" || exit 1
local/prepare_en_transcription.sh $WORK/local $WORK/local/dict || exit 1
local/create_phone_lists.sh $WORK/local/dict || exit 1
utils/prepare_lang.sh $WORK/local/dict '_SIL_' $WORK/local/lang $WORK/lang || exit 1
local/create_G.sh $WORK/lang "$LMs" $WORK/local/lm $WORK/local/dict/lexicon.txt || exit 1
echo "Create MFCC features and storing them (Could be large)."
for s in train $TEST_SETS ; do
steps/make_mfcc.sh --mfcc-config common/mfcc.conf --cmd \
"$train_cmd" --nj $njobs $WORK/local/$s $EXP/make_mfcc/$s $WORK/mfcc || exit 1;
# Note --fake -> NO CMVN
steps/compute_cmvn_stats.sh $fake $WORK/local/$s \
$EXP/make_mfcc/$s $WORK/mfcc || exit 1;
done
echo "Decoding is done for each pair (TEST_SET x LMs)"
echo "Distribute the links to MFCC feats to all LM variations."
cp $WORK/local/train/feats.scp $WORK/train/feats.scp
cp $WORK/local/train/cmvn.scp $WORK/train/cmvn.scp
for s in $TEST_SETS; do
for lm in $LMs; do
tgt_dir=${s}_`basename "$lm"`
echo "cp $WORK/local/$s/feats.scp $WORK/$tgt_dir/feats.scp"
cp $WORK/local/$s/feats.scp $WORK/$tgt_dir/feats.scp
echo "cp $WORK/local/$s/cmvn.scp $WORK/$tgt_dir/cmvn.scp"
cp $WORK/local/$s/cmvn.scp $WORK/$tgt_dir/cmvn.scp
done
done
#######################################################################
# Training Acoustic Models #
#######################################################################
echo "Train monophone models on full data -> may be wastefull (can be done on subset)"
steps/train_mono.sh --nj $njobs --cmd "$train_cmd" $WORK/train $WORK/lang $EXP/mono || exit 1;
echo "Get alignments from monophone system."
steps/align_si.sh --nj $njobs --cmd "$train_cmd" \
$WORK/train $WORK/lang $EXP/mono $EXP/mono_ali || exit 1;
echo "Train tri1 [first triphone pass]"
steps/train_deltas.sh --cmd "$train_cmd" \
$pdf $gauss $WORK/train $WORK/lang $EXP/mono_ali $EXP/tri1 || exit 1;
# draw-tree $WORK/lang/phones.txt $EXP/tri1/tree | dot -Tsvg -Gsize=8,10.5 > graph.svg
echo "Align tri1"
steps/align_si.sh --nj $njobs --cmd "$train_cmd" \
--use-graphs true $WORK/train $WORK/lang $EXP/tri1 $EXP/tri1_ali || exit 1;
echo "Train tri2a [delta+delta-deltas]"
steps/train_deltas.sh --cmd "$train_cmd" $pdf $gauss \
$WORK/train $WORK/lang $EXP/tri1_ali $EXP/tri2a || exit 1;
echo "Train tri2b [LDA+MLLT]"
steps/train_lda_mllt.sh --cmd "$train_cmd" $pdf $gauss \
$WORK/train $WORK/lang $EXP/tri1_ali $EXP/tri2b || exit 1;
echo "Align all data with LDA+MLLT system (tri2b)"
steps/align_si.sh --nj $njobs --cmd "$train_cmd" \
--use-graphs true $WORK/train $WORK/lang $EXP/tri2b $EXP/tri2b_ali || exit 1;
echo "Train MMI on top of LDA+MLLT."
steps/make_denlats.sh --nj $njobs --cmd "$train_cmd" \
--beam $mmi_beam --lattice-beam $mmi_lat_beam \
$WORK/train $WORK/lang $EXP/tri2b $EXP/tri2b_denlats || exit 1;
steps/train_mmi.sh $WORK/train $WORK/lang $EXP/tri2b_ali $EXP/tri2b_denlats $EXP/tri2b_mmi || exit 1;
echo "Train MMI on top of LDA+MLLT with boosting. train_mmi_boost is a e.g. 0.05"
steps/train_mmi.sh --boost ${train_mmi_boost} $WORK/train $WORK/lang \
$EXP/tri2b_ali $EXP/tri2b_denlats $EXP/tri2b_mmi_b${train_mmi_boost} || exit 1;
echo "Train MPE."
steps/train_mpe.sh $WORK/train $WORK/lang $EXP/tri2b_ali $EXP/tri2b_denlats $EXP/tri2b_mpe || exit 1;
#######################################################################
# Building decoding graph #
#######################################################################
for lm in $LMs ; do
lm=`basename "$lm"`
utils/mkgraph.sh $WORK/lang_${lm} $EXP/mono $EXP/mono/graph_${lm} || exit 1
utils/mkgraph.sh $WORK/lang_${lm} $EXP/tri1 $EXP/tri1/graph_${lm} || exit 1
utils/mkgraph.sh $WORK/lang_${lm} $EXP/tri2a $EXP/tri2a/graph_${lm} || exit 1
utils/mkgraph.sh $WORK/lang_${lm} $EXP/tri2b $EXP/tri2b/graph_${lm} || exit 1
done
#######################################################################
# Decoding #
#######################################################################
for s in $TEST_SETS ; do
for lm in $LMs ; do
lm=`basename "$lm"`
tgt_dir=${s}_`basename "$lm"`
echo "Monophone decoding"
# Note: steps/decode.sh --scoring-opts "--min-lmw $min_lmw --max-lmw $max_lmw" \
# calls the command line once for each test,
# and afterwards averages the WERs into (in this case $EXP/mono/decode/)
steps/decode.sh --scoring-opts "--min-lmw $min_lmw --max-lmw $max_lmw" \
--config common/decode.conf --nj $njobs --cmd "$decode_cmd" \
$EXP/mono/graph_${lm} $WORK/${tgt_dir} $EXP/mono/decode_${tgt_dir}
echo "Decode tri1"
steps/decode.sh --scoring-opts "--min-lmw $min_lmw --max-lmw $max_lmw" \
--config common/decode.conf --nj $njobs --cmd "$decode_cmd" \
$EXP/tri1/graph_${lm} $WORK/$tgt_dir $EXP/tri1/decode_${tgt_dir}
echo "Decode tri2a"
steps/decode.sh --scoring-opts "--min-lmw $min_lmw --max-lmw $max_lmw" \
--config common/decode.conf --nj $njobs --cmd "$decode_cmd" \
$EXP/tri2a/graph_${lm} $WORK/$tgt_dir $EXP/tri2a/decode_${tgt_dir}
echo "Decode tri2b [LDA+MLLT]"
steps/decode.sh --scoring-opts "--min-lmw $min_lmw --max-lmw $max_lmw" \
--config common/decode.conf --nj $njobs --cmd "$decode_cmd" \
$EXP/tri2b/graph_${lm} $WORK/$tgt_dir $EXP/tri2b/decode_${tgt_dir}
# Note: change --iter option to select the best model. 4.mdl == final.mdl
echo "Decode MMI on top of LDA+MLLT."
steps/decode.sh --scoring-opts "--min-lmw $min_lmw --max-lmw $max_lmw" \
--config common/decode.conf --iter 4 --nj $njobs --cmd "$decode_cmd" \
$EXP/tri2b/graph_${lm} $WORK/$tgt_dir $EXP/tri2b_mmi/decode_it4_${tgt_dir}
echo "Decode MMI on top of LDA+MLLT with boosting. train_mmi_boost is a number e.g. 0.05"
steps/decode.sh --scoring-opts "--min-lmw $min_lmw --max-lmw $max_lmw" \
--config common/decode.conf --iter 4 --nj $njobs --cmd "$decode_cmd" \
$EXP/tri2b/graph_${lm} $WORK/$tgt_dir $EXP/tri2b_mmi_b${train_mmi_boost}/decode_it4_${tgt_dir};
echo "Decode MPE."
steps/decode.sh --scoring-opts "--min-lmw $min_lmw --max-lmw $max_lmw" \
--config common/decode.conf --iter 4 --nj $njobs --cmd "$decode_cmd" \
$EXP/tri2b/graph_${lm} $WORK/$tgt_dir $EXP/tri2b_mpe/decode_it4_${tgt_dir} || exit 1;
done
done
echo "Successfully trained and evaluated all the experiments"
local/results.py $EXP | tee $EXP/results.log
local/export_models.sh $TGT_MODELS $EXP $WORK/lang
|