Adding orth-factorized-TDNN for fisher_swbd

egs/fisher_swbd/s5/local/chain/run_tdnn_7c.sh

@@ -0,0 +1,273 @@
+#!/bin/bash
+# Copyright 2017 University of Chinese Academy of Sciences (UCAS) Gaofeng Cheng
+# Apache 2.0
+
+# Based on tdnn_7m (from egs/swbd/s5c).
+
+# But tdnn_7c enjoys two more TDNN layers to make [-21,21] input temporal context,
+# this is intended as an extra baseline for 7d.
+
+# System               tdnn_7b_sp tdnn_7c_sp 
+# WER on eval2000(tg)    13.6     13.5    
+# WER on eval2000(fg)    13.3     13.3    
+# WER on rt03(tg)        12.7     12.7    
+# WER on rt03(fg)        12.4     12.5    
+# Final train prob                -0.103
+# Final valid prob                -0.114
+# Final train prob (xent)         -1.159
+# Final valid prob (xent)         -1.2024
+# Num-parameters                  18781673
+
+# %WER 17.8 | 2628 21594 | 84.1 9.3 6.5 1.9 17.8 56.5 | exp/chain/tdnn_7c_sp/decode_eval2000_fsh_sw1_tg/score_8_0.0/eval2000_hires.ctm.callhm.filt.sys
+# %WER 13.5 | 4459 42989 | 88.1 7.5 4.4 1.6 13.5 51.3 | exp/chain/tdnn_7c_sp/decode_eval2000_fsh_sw1_tg/score_8_0.0/eval2000_hires.ctm.filt.sys
+# %WER 9.2 | 1831 21395 | 92.1 5.6 2.3 1.3 9.2 44.0 | exp/chain/tdnn_7c_sp/decode_eval2000_fsh_sw1_tg/score_8_0.0/eval2000_hires.ctm.swbd.filt.sys
+# %WER 17.7 | 2628 21594 | 84.3 9.2 6.5 2.0 17.7 55.9 | exp/chain/tdnn_7c_sp/decode_eval2000_fsh_sw1_fg/score_8_0.0/eval2000_hires.ctm.callhm.filt.sys
+# %WER 13.3 | 4459 42989 | 88.3 7.4 4.3 1.6 13.3 50.5 | exp/chain/tdnn_7c_sp/decode_eval2000_fsh_sw1_fg/score_8_0.0/eval2000_hires.ctm.filt.sys
+# %WER 8.9 | 1831 21395 | 92.2 5.1 2.8 1.0 8.9 42.4 | exp/chain/tdnn_7c_sp/decode_eval2000_fsh_sw1_fg/score_10_0.0/eval2000_hires.ctm.swbd.filt.sys
+
+# %WER 10.2 | 3970 36721 | 90.8 5.7 3.5 1.1 10.2 42.0 | exp/chain/tdnn_7c_sp/decode_rt03_fsh_sw1_tg/score_8_0.0/rt03_hires.ctm.fsh.filt.sys
+# %WER 12.7 | 8420 76157 | 88.6 7.2 4.2 1.3 12.7 44.3 | exp/chain/tdnn_7c_sp/decode_rt03_fsh_sw1_tg/score_8_0.0/rt03_hires.ctm.filt.sys
+# %WER 15.0 | 4450 39436 | 86.5 8.6 4.9 1.5 15.0 46.4 | exp/chain/tdnn_7c_sp/decode_rt03_fsh_sw1_tg/score_8_0.0/rt03_hires.ctm.swbd.filt.sys
+# %WER 9.9 | 3970 36721 | 91.1 5.4 3.5 1.0 9.9 41.6 | exp/chain/tdnn_7c_sp/decode_rt03_fsh_sw1_fg/score_8_0.0/rt03_hires.ctm.fsh.filt.sys
+# %WER 12.5 | 8420 76157 | 88.8 7.0 4.2 1.3 12.5 43.9 | exp/chain/tdnn_7c_sp/decode_rt03_fsh_sw1_fg/score_8_0.0/rt03_hires.ctm.filt.sys
+# %WER 14.8 | 4450 39436 | 86.6 7.9 5.6 1.4 14.8 46.1 | exp/chain/tdnn_7c_sp/decode_rt03_fsh_sw1_fg/score_9_0.0/rt03_hires.ctm.swbd.filt.sys
+
+
+set -e
+
+# configs for 'chain'
+stage=12
+train_stage=-10
+get_egs_stage=-10
+speed_perturb=true
+dir=exp/chain/tdnn_7c # Note: _sp will get added to this if $speed_perturb == true.
+decode_iter=
+decode_dir_affix=
+
+# training options
+leftmost_questions_truncate=-1
+num_epochs=4
+initial_effective_lrate=0.001
+final_effective_lrate=0.0001
+max_param_change=2.0
+num_jobs_initial=3
+num_jobs_final=16
+minibatch_size=128
+frames_per_eg=150
+remove_egs=false
+common_egs_dir=
+xent_regularize=0.1
+
+# End configuration section.
+echo "$0 $@"  # Print the command line for logging
+
+. ./cmd.sh
+. ./path.sh
+. ./utils/parse_options.sh
+
+if ! cuda-compiled; then
+  cat <<EOF && exit 1
+This script is intended to be used with GPUs but you have not compiled Kaldi with CUDA
+If you want to use GPUs (and have them), go to src/, and configure and make on a machine
+where "nvcc" is installed.
+EOF
+fi
+
+# The iVector-extraction and feature-dumping parts are the same as the standard
+# nnet3 setup, and you can skip them by setting "--stage 8" if you have already
+# run those things.
+
+suffix=
+if [ "$speed_perturb" == "true" ]; then
+  suffix=_sp
+fi
+
+dir=${dir}$suffix
+build_tree_train_set=train_nodup
+train_set=train_nodup_sp
+build_tree_ali_dir=exp/tri5a_ali
+treedir=exp/chain/tri6_tree
+lang=data/lang_chain
+
+# if we are using the speed-perturbed data we need to generate
+# alignments for it.
+local/nnet3/run_ivector_common.sh --stage $stage \
+  --speed-perturb $speed_perturb \
+  --generate-alignments $speed_perturb || exit 1;
+
+if [ $stage -le 9 ]; then
+  # Get the alignments as lattices (gives the CTC training more freedom).
+  # use the same num-jobs as the alignments
+  nj=$(cat $build_tree_ali_dir/num_jobs) || exit 1;
+  steps/align_fmllr_lats.sh --nj $nj --cmd "$train_cmd" data/$train_set \
+    data/lang exp/tri5a exp/tri5a_lats_nodup$suffix
+  rm exp/tri5a_lats_nodup$suffix/fsts.*.gz # save space
+fi
+
+if [ $stage -le 10 ]; 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;
+  # Use our special topology... note that later on may have to tune this
+  # topology.
+  steps/nnet3/chain/gen_topo.py $nonsilphonelist $silphonelist >$lang/topo
+fi
+
+if [ $stage -le 11 ]; then
+  # Build a tree using our new topology.
+  steps/nnet3/chain/build_tree.sh --frame-subsampling-factor 3 \
+      --leftmost-questions-truncate $leftmost_questions_truncate \
+      --context-opts "--context-width=2 --central-position=1" \
+      --cmd "$train_cmd" 11000 data/$build_tree_train_set $lang $build_tree_ali_dir $treedir
+fi
+
+if [ $stage -le 12 ]; then
+  echo "$0: creating neural net configs using the xconfig parser";
+
+  num_targets=$(tree-info $treedir/tree |grep num-pdfs|awk '{print $2}')
+  learning_rate_factor=$(echo "print 0.5/$xent_regularize" | python)
+
+  mkdir -p $dir/configs
+  cat <<EOF > $dir/configs/network.xconfig
+  input dim=100 name=ivector
+  input dim=40 name=input
+
+  # please note that it is important to have input layer with the name=input
+  # as the layer immediately preceding the fixed-affine-layer to enable
+  # the use of short notation for the descriptor
+  fixed-affine-layer name=lda input=Append(-1,0,1,ReplaceIndex(ivector, t, 0)) affine-transform-file=$dir/configs/lda.mat
+
+  # the first splicing is moved before the lda layer, so no splicing here
+  relu-batchnorm-layer name=tdnn1 dim=625
+  relu-batchnorm-layer name=tdnn2 input=Append(-1,0,1) dim=625
+  relu-batchnorm-layer name=tdnn3 dim=625
+  relu-batchnorm-layer name=tdnn4 input=Append(-1,0,1) dim=625
+  relu-batchnorm-layer name=tdnn5 dim=625
+  relu-batchnorm-layer name=tdnn6 input=Append(-3,0,3) dim=625
+  relu-batchnorm-layer name=tdnn7 input=Append(-3,0,3) dim=625
+  relu-batchnorm-layer name=tdnn8 input=Append(-3,0,3) dim=625
+  relu-batchnorm-layer name=tdnn9 input=Append(-3,0,3) dim=625
+  relu-batchnorm-layer name=tdnn10 input=Append(-3,0,3) dim=625
+  relu-batchnorm-layer name=tdnn11 input=Append(-3,0,3) dim=625
+
+  ## adding the layers for chain branch
+  relu-batchnorm-layer name=prefinal-chain input=tdnn11 dim=625 target-rms=0.5
+  output-layer name=output include-log-softmax=false dim=$num_targets max-change=1.5
+
+  # adding the layers for xent branch
+  # This block prints the configs for a separate output that will be
+  # trained with a cross-entropy objective in the 'chain' models... this
+  # has the effect of regularizing the hidden parts of the model.  we use
+  # 0.5 / args.xent_regularize as the learning rate factor- the factor of
+  # 0.5 / args.xent_regularize is suitable as it means the xent
+  # final-layer learns at a rate independent of the regularization
+  # constant; and the 0.5 was tuned so as to make the relative progress
+  # similar in the xent and regular final layers.
+  relu-batchnorm-layer name=prefinal-xent input=tdnn11 dim=625 target-rms=0.5
+  output-layer name=output-xent dim=$num_targets learning-rate-factor=$learning_rate_factor max-change=1.5
+
+EOF
+  steps/nnet3/xconfig_to_configs.py --xconfig-file $dir/configs/network.xconfig --config-dir $dir/configs/
+fi
+
+if [ $stage -le 13 ]; then
+  if [[ $(hostname -f) == *.clsp.jhu.edu ]] && [ ! -d $dir/egs/storage ]; then
+    utils/create_split_dir.pl \
+     /export/b0{5,6,7,8}/$USER/kaldi-data/egs/swbd-$(date +'%m_%d_%H_%M')/s5c/$dir/egs/storage $dir/egs/storage
+  fi
+
+  steps/nnet3/chain/train.py --stage $train_stage \
+    --cmd "$decode_cmd" \
+    --feat.online-ivector-dir exp/nnet3/ivectors_${train_set} \
+    --feat.cmvn-opts "--norm-means=false --norm-vars=false" \
+    --chain.xent-regularize $xent_regularize \
+    --chain.leaky-hmm-coefficient 0.1 \
+    --chain.l2-regularize 0.00005 \
+    --chain.apply-deriv-weights false \
+    --chain.lm-opts="--num-extra-lm-states=2000" \
+    --egs.dir "$common_egs_dir" \
+    --egs.stage $get_egs_stage \
+    --egs.opts "--frames-overlap-per-eg 0" \
+    --egs.chunk-width $frames_per_eg \
+    --trainer.num-chunk-per-minibatch $minibatch_size \
+    --trainer.frames-per-iter 1500000 \
+    --trainer.num-epochs $num_epochs \
+    --trainer.optimization.num-jobs-initial $num_jobs_initial \
+    --trainer.optimization.num-jobs-final $num_jobs_final \
+    --trainer.optimization.initial-effective-lrate $initial_effective_lrate \
+    --trainer.optimization.final-effective-lrate $final_effective_lrate \
+    --trainer.max-param-change $max_param_change \
+    --cleanup.remove-egs $remove_egs \
+    --feat-dir data/${train_set}_hires \
+    --tree-dir $treedir \
+    --lat-dir exp/tri5a_lats_nodup$suffix \
+    --dir $dir  || exit 1;
+fi
+
+if [ $stage -le 14 ]; then
+  # Note: it might appear that this $lang directory is mismatched, and it is as
+  # far as the 'topo' is concerned, but this script doesn't read the 'topo' from
+  # the lang directory.
+  utils/mkgraph.sh --self-loop-scale 1.0 data/lang_fsh_sw1_tg $dir $dir/graph_fsh_sw1_tg
+fi
+
+decode_suff=fsh_sw1_tg
+graph_dir=$dir/graph_fsh_sw1_tg
+if [ $stage -le 15 ]; then
+  if [ ! -z $decode_iter ]; then
+    iter_opts=" --iter $decode_iter "
+  fi
+  for decode_set in rt03 eval2000; do
+      (
+      steps/nnet3/decode.sh --acwt 1.0 --post-decode-acwt 10.0 \
+          --nj 50 --cmd "$decode_cmd" $iter_opts \
+          --online-ivector-dir exp/nnet3/ivectors_${decode_set} \
+         $graph_dir data/${decode_set}_hires \
+         $dir/decode_${decode_set}${decode_dir_affix:+_$decode_dir_affix}_${decode_suff} || exit 1;
+      if $has_fisher; then
+          steps/lmrescore_const_arpa.sh --cmd "$decode_cmd" \
+            data/lang_fsh_sw1_{tg,fg} data/${decode_set}_hires \
+            $dir/decode_${decode_set}${decode_dir_affix:+_$decode_dir_affix}_fsh_sw1_{tg,fg} || exit 1;
+      fi
+      ) &
+  done
+fi
+
+test_online_decoding=true
+lang=data/lang_fsh_sw1_tg
+if $test_online_decoding && [ $stage -le 16 ]; then
+  # note: if the features change (e.g. you add pitch features), you will have to
+  # change the options of the following command line.
+  steps/online/nnet3/prepare_online_decoding.sh \
+       --mfcc-config conf/mfcc_hires.conf \
+       $lang exp/nnet3/extractor $dir ${dir}_online
+
+  rm $dir/.error 2>/dev/null || true
+  for decode_set in rt03 eval2000; do
+    (
+      # note: we just give it "$decode_set" as it only uses the wav.scp, the
+      # feature type does not matter.
+
+      steps/online/nnet3/decode.sh --nj 50 --cmd "$decode_cmd" $iter_opts \
+          --acwt 1.0 --post-decode-acwt 10.0 \
+         $graph_dir data/${decode_set}_hires \
+         ${dir}_online/decode_${decode_set}${decode_iter:+_$decode_iter}_${decode_suff} || exit 1;
+      if $has_fisher; then
+	      steps/lmrescore_const_arpa.sh --cmd "$decode_cmd" \
+		      data/lang_fsh_sw1_{tg,fg} data/${decode_set}_hires \
+		      ${dir}_online/decode_${decode_set}${decode_dir_affix:+_$decode_dir_affix}_fsh_sw1_{tg,fg} || exit 1;
+      fi
+    ) || touch $dir/.error &
+  done
+  wait
+  if [ -f $dir/.error ]; then
+    echo "$0: something went wrong in online decoding"
+    exit 1
+  fi
+fi
+
+exit 0;