-
Notifications
You must be signed in to change notification settings - Fork 5.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Adding scripts for yomdle Russian #2953
Merged
Merged
Changes from 3 commits
Commits
Show all changes
5 commits
Select commit
Hold shift + click to select a range
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,3 @@ | ||
This directory contains example scripts for OCR on the Yomdle and Slam datasets. | ||
Training is done on the Yomdle dataset and testing is done on Slam. | ||
LM rescoring is also done with extra corpus data obtained from various sources |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,12 @@ | ||
# you can change cmd.sh depending on what type of queue you are using. | ||
# If you have no queueing system and want to run on a local machine, you | ||
# can change all instances 'queue.pl' to run.pl (but be careful and run | ||
# commands one by one: most recipes will exhaust the memory on your | ||
# machine). queue.pl works with GridEngine (qsub). slurm.pl works | ||
# with slurm. Different queues are configured differently, with different | ||
# queue names and different ways of specifying things like memory; | ||
# to account for these differences you can create and edit the file | ||
# conf/queue.conf to match your queue's configuration. Search for | ||
# conf/queue.conf in https://kaldi-asr.org/doc/queue.html for more information, | ||
# or search for the string 'default_config' in utils/queue.pl or utils/slurm.pl. | ||
export cmd="queue.pl" |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1 @@ | ||
../../cifar/v1/image/ |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,36 @@ | ||
#!/bin/bash | ||
# Copyright 2018 Hossein Hadian | ||
# 2018 Ashish Arora | ||
|
||
# Apache 2.0 | ||
# This script performs data augmentation. | ||
|
||
nj=4 | ||
cmd=run.pl | ||
feat_dim=40 | ||
verticle_shift=0 | ||
echo "$0 $@" | ||
|
||
. ./cmd.sh | ||
. ./path.sh | ||
. ./utils/parse_options.sh || exit 1; | ||
|
||
srcdir=$1 | ||
outdir=$2 | ||
datadir=$3 | ||
|
||
mkdir -p $datadir/augmentations | ||
echo "copying $srcdir to $datadir/augmentations/aug1, allowed length, creating feats.scp" | ||
|
||
for set in aug1; do | ||
image/copy_data_dir.sh --spk-prefix $set- --utt-prefix $set- \ | ||
$srcdir $datadir/augmentations/$set | ||
cat $srcdir/allowed_lengths.txt > $datadir/augmentations/$set/allowed_lengths.txt | ||
local/extract_features.sh --nj $nj --cmd "$cmd" --feat-dim $feat_dim \ | ||
--vertical-shift $verticle_shift \ | ||
--fliplr false --augment 'random_scale' $datadir/augmentations/$set | ||
done | ||
|
||
echo " combine original data and data from different augmentations" | ||
utils/combine_data.sh --extra-files images.scp $outdir $srcdir $datadir/augmentations/aug1 | ||
cat $srcdir/allowed_lengths.txt > $outdir/allowed_lengths.txt |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,66 @@ | ||
#!/bin/bash | ||
|
||
# this script is used for comparing decoding results between systems. | ||
# e.g. local/chain/compare_wer.sh exp/chain/cnn{1a,1b} | ||
|
||
# Copyright 2017 Chun Chieh Chang | ||
# 2017 Ashish Arora | ||
|
||
if [ $# == 0 ]; then | ||
echo "Usage: $0: <dir1> [<dir2> ... ]" | ||
echo "e.g.: $0 exp/chain/cnn{1a,1b}" | ||
exit 1 | ||
fi | ||
|
||
echo "# $0 $*" | ||
used_epochs=false | ||
|
||
echo -n "# System " | ||
for x in $*; do printf "% 10s" " $(basename $x)"; done | ||
echo | ||
|
||
echo -n "# WER " | ||
for x in $*; do | ||
wer=$(cat $x/decode_test/scoring_kaldi/best_wer | awk '{print $2}') | ||
printf "% 10s" $wer | ||
done | ||
echo | ||
|
||
echo -n "# WER (rescored) " | ||
for x in $*; do | ||
wer=$(cat $x/decode_test_rescored/scoring_kaldi/best_wer | awk '{print $2}') | ||
printf "% 10s" $wer | ||
done | ||
echo | ||
|
||
echo -n "# CER " | ||
for x in $*; do | ||
cer=$(cat $x/decode_test/scoring_kaldi/best_cer | awk '{print $2}') | ||
printf "% 10s" $cer | ||
done | ||
echo | ||
|
||
echo -n "# CER (rescored) " | ||
for x in $*; do | ||
cer=$(cat $x/decode_test_rescored/scoring_kaldi/best_cer | awk '{print $2}') | ||
printf "% 10s" $cer | ||
done | ||
echo | ||
|
||
if $used_epochs; then | ||
exit 0; # the diagnostics aren't comparable between regular and discriminatively trained systems. | ||
fi | ||
|
||
echo -n "# Final train prob " | ||
for x in $*; do | ||
prob=$(grep Overall $x/log/compute_prob_train.final.log | grep -v xent | awk '{printf("%.4f", $8)}') | ||
printf "% 10s" $prob | ||
done | ||
echo | ||
|
||
echo -n "# Final valid prob " | ||
for x in $*; do | ||
prob=$(grep Overall $x/log/compute_prob_valid.final.log | grep -v xent | awk '{printf("%.4f", $8)}') | ||
printf "% 10s" $prob | ||
done | ||
echo |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1 @@ | ||
tuning/run_cnn_e2eali_1a.sh |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,131 @@ | ||
#!/bin/bash | ||
|
||
# Copyright 2017 Hossein Hadian | ||
# This script does end2end chain training (i.e. from scratch) | ||
# local/chain/compare_wer.sh exp/chain/e2e_cnn_1a/ | ||
# System e2e_cnn_1a | ||
# score_basic rescoring + nomalized | ||
# WER 16.24 11.0 | ||
# WER (rescored) 15.63 10.5 | ||
# CER 5.98 5.6 | ||
# CER (rescored) 5.66 5.3 | ||
# Final train prob 0.1376 | ||
# Final valid prob 0.1913 | ||
# steps/info/chain_dir_info.pl exp/chain/e2e_cnn_1a | ||
# exp/chain/e2e_cnn_1a: num-iters=27 nj=5..8 num-params=3.0M dim=40->470 combine=0.091->0.091 (over 1) logprob:train/valid[17,26,final]=(0.135,0.137,0.138/0.191,0.191,0.191) | ||
|
||
set -e | ||
# configs for 'chain' | ||
stage=0 | ||
nj=30 | ||
train_stage=-10 | ||
get_egs_stage=-10 | ||
affix=1a | ||
|
||
# training options | ||
tdnn_dim=450 | ||
minibatch_size=150=64,32/300=32,16/600=16,8/1200=8,4 | ||
cmvn_opts="--norm-means=false --norm-vars=false" | ||
train_set=train | ||
lang_decode=data/lang | ||
decode_e2e=true | ||
# End configuration section. | ||
echo "$0 $@" # Print the command line for logging | ||
|
||
. ./cmd.sh | ||
. ./path.sh | ||
. ./utils/parse_options.sh | ||
|
||
if ! cuda-compiled; then | ||
cat <<EOF && exit 1 | ||
This script is intended to be used with GPUs but you have not compiled Kaldi with CUDA | ||
If you want to use GPUs (and have them), go to src/, and configure and make on a machine | ||
where "nvcc" is installed. | ||
EOF | ||
fi | ||
|
||
lang=data/lang_e2e | ||
treedir=exp/chain/e2e_monotree # it's actually just a trivial tree (no tree building) | ||
dir=exp/chain/e2e_cnn_${affix} | ||
|
||
if [ $stage -le 0 ]; then | ||
# Create a version of the lang/ directory that has one state per phone in the | ||
# topo file. [note, it really has two states.. the first one is only repeated | ||
# once, the second one has zero or more repeats.] | ||
rm -rf $lang | ||
cp -r data/lang $lang | ||
silphonelist=$(cat $lang/phones/silence.csl) || exit 1; | ||
nonsilphonelist=$(cat $lang/phones/nonsilence.csl) || exit 1; | ||
steps/nnet3/chain/gen_topo.py $nonsilphonelist $silphonelist >$lang/topo | ||
fi | ||
|
||
if [ $stage -le 1 ]; then | ||
steps/nnet3/chain/e2e/prepare_e2e.sh --nj $nj --cmd "$cmd" \ | ||
--shared-phones true \ | ||
--type mono \ | ||
data/$train_set $lang $treedir | ||
$cmd $treedir/log/make_phone_lm.log \ | ||
cat data/$train_set/text \| \ | ||
steps/nnet3/chain/e2e/text_to_phones.py data/lang \| \ | ||
utils/sym2int.pl -f 2- data/lang/phones.txt \| \ | ||
chain-est-phone-lm --num-extra-lm-states=500 \ | ||
ark:- $treedir/phone_lm.fst | ||
fi | ||
|
||
if [ $stage -le 2 ]; then | ||
echo "$0: creating neural net configs using the xconfig parser"; | ||
num_targets=$(tree-info $treedir/tree | grep num-pdfs | awk '{print $2}') | ||
cnn_opts="l2-regularize=0.075" | ||
tdnn_opts="l2-regularize=0.075" | ||
output_opts="l2-regularize=0.1" | ||
common1="$cnn_opts required-time-offsets= height-offsets=-2,-1,0,1,2 num-filters-out=36" | ||
common2="$cnn_opts required-time-offsets= height-offsets=-2,-1,0,1,2 num-filters-out=70" | ||
common3="$cnn_opts required-time-offsets= height-offsets=-1,0,1 num-filters-out=70" | ||
|
||
mkdir -p $dir/configs | ||
cat <<EOF > $dir/configs/network.xconfig | ||
input dim=40 name=input | ||
conv-relu-batchnorm-layer name=cnn1 height-in=40 height-out=40 time-offsets=-3,-2,-1,0,1,2,3 $common1 | ||
conv-relu-batchnorm-layer name=cnn2 height-in=40 height-out=20 time-offsets=-2,-1,0,1,2 $common1 height-subsample-out=2 | ||
conv-relu-batchnorm-layer name=cnn3 height-in=20 height-out=20 time-offsets=-4,-2,0,2,4 $common2 | ||
conv-relu-batchnorm-layer name=cnn4 height-in=20 height-out=20 time-offsets=-4,-2,0,2,4 $common2 | ||
conv-relu-batchnorm-layer name=cnn5 height-in=20 height-out=10 time-offsets=-4,-2,0,2,4 $common2 height-subsample-out=2 | ||
conv-relu-batchnorm-layer name=cnn6 height-in=10 height-out=10 time-offsets=-4,0,4 $common3 | ||
conv-relu-batchnorm-layer name=cnn7 height-in=10 height-out=10 time-offsets=-4,0,4 $common3 | ||
relu-batchnorm-layer name=tdnn1 input=Append(-4,0,4) dim=$tdnn_dim $tdnn_opts | ||
relu-batchnorm-layer name=tdnn2 input=Append(-4,0,4) dim=$tdnn_dim $tdnn_opts | ||
relu-batchnorm-layer name=tdnn3 input=Append(-4,0,4) dim=$tdnn_dim $tdnn_opts | ||
## adding the layers for chain branch | ||
relu-batchnorm-layer name=prefinal-chain dim=$tdnn_dim target-rms=0.5 $output_opts | ||
output-layer name=output include-log-softmax=false dim=$num_targets max-change=1.5 $output_opts | ||
EOF | ||
|
||
steps/nnet3/xconfig_to_configs.py --xconfig-file $dir/configs/network.xconfig --config-dir $dir/configs | ||
fi | ||
|
||
if [ $stage -le 3 ]; then | ||
steps/nnet3/chain/e2e/train_e2e.py --stage $train_stage \ | ||
--cmd "$cmd" \ | ||
--feat.cmvn-opts "$cmvn_opts" \ | ||
--chain.leaky-hmm-coefficient 0.1 \ | ||
--chain.apply-deriv-weights true \ | ||
--egs.stage $get_egs_stage \ | ||
--egs.opts "--num_egs_diagnostic 100 --num_utts_subset 400" \ | ||
--chain.frame-subsampling-factor 4 \ | ||
--chain.alignment-subsampling-factor 4 \ | ||
--trainer.add-option="--optimization.memory-compression-level=2" \ | ||
--trainer.num-chunk-per-minibatch $minibatch_size \ | ||
--trainer.frames-per-iter 1500000 \ | ||
--trainer.num-epochs 3 \ | ||
--trainer.optimization.momentum 0 \ | ||
--trainer.optimization.num-jobs-initial 5 \ | ||
--trainer.optimization.num-jobs-final 8 \ | ||
--trainer.optimization.initial-effective-lrate 0.001 \ | ||
--trainer.optimization.final-effective-lrate 0.0001 \ | ||
--trainer.optimization.shrink-value 1.0 \ | ||
--trainer.max-param-change 2.0 \ | ||
--cleanup.remove-egs true \ | ||
--feat-dir data/${train_set} \ | ||
--tree-dir $treedir \ | ||
--dir $dir || exit 1; | ||
fi |
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
echo statements should usually start with "$0: " so that it's clear which script produced them.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
sorry, local/augment_data.sh is not used yet for yomdle_russian setup. I will remove it.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
oh, ok, I got it now... making changes.