Made changes to the augmentation script to make it work for ASR and speaker ID

egs/swbd/s5c/local/chain/multi_style/tuning/run_tdnn_1a.sh

@@ -0,0 +1,255 @@
+#!/bin/bash
+
+# This recipe does multi-style training of TDNN model
+
+set -e
+
+# configs for 'chain'
+stage=0
+train_stage=-10
+get_egs_stage=-10
+num_epochs=3
+
+# Augmentation options
+augmentation_list="reverb:babble:music:noise:clean"
+use_ivectors=true
+
+affix=1a
+suffix="_ms"
+if [ -e data/rt03 ]; then maybe_rt03=rt03; else maybe_rt03= ; fi
+
+decode_iter=
+decode_nj=50
+
+# training options
+frames_per_eg=150,110,100
+remove_egs=false
+common_egs_dir=
+xent_regularize=0.1
+dropout_schedule='0,[email protected],[email protected],0'
+
+test_online_decoding=false  # if true, it will run the last decoding stage.
+
+# End configuration section.
+echo "$0 $@"  # Print the command line for logging
+
+. ./cmd.sh
+. ./path.sh
+. ./utils/parse_options.sh
+
+dir=exp/chain/tdnn${affix}${suffix}
+
+if ! cuda-compiled; then
+  cat <<EOF && exit 1
+This script is intended to be used with GPUs but you have not compiled Kaldi with CUDA
+If you want to use GPUs (and have them), go to src/, and configure and make on a machine
+where "nvcc" is installed.
+EOF
+fi
+
+# The iVector-extraction and feature-dumping parts are the same as the standard
+# nnet3 setup, and you can skip them by setting "--stage 8" if you have already
+# run those things.
+
+clean_set=train_nodup
+clean_ali=tri4_ali_nodup
+train_set=$clean_set$suffix # Will be prepared by the script local/nnet3/prepare_multistyle_data.sh
+ali_dir=$clean_ali$suffix
+treedir=exp/chain/tri5_7d_tree$suffix
+lang=data/lang_chain_2y
+
+
+# if we are using the speed-perturbed data we need to generate
+# alignments for it.
+local/nnet3/prepare_multistyle_data.sh --stage $stage \
+  --augmentation-list "$augmentation_list" \
+  --use-ivectors "$use_ivectors" \
+  --train-set $clean_set --clean-ali $clean_ali || exit 1;
+
+if [ $stage -le 11 ]; then
+  # Get the alignments as lattices (gives the LF-MMI training more freedom).
+  # use the same num-jobs as the alignments
+  nj=$(cat exp/tri4_ali_nodup$suffix/num_jobs) || exit 1;
+  steps/align_fmllr_lats.sh --nj $nj --cmd "$train_cmd" data/${clean_set} \
+    data/lang exp/tri4 exp/tri4_lats_nodup${suffix}_clean
+  rm exp/tri4_lats_nodup${suffix}_clean/fsts.*.gz # save space
+  local/copy_lat_dir.sh --nj $nj --cmd "$train_cmd" \
+    data/${train_set} exp/tri4_lats_nodup${suffix}_clean exp/tri4_lats_nodup${suffix} || exit 1;
+fi
+
+if [ $stage -le 12 ]; 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 [ $stage -le 13 ]; then
+  # Build a tree using our new topology. This is the critically different
+  # step compared with other recipes.
+  steps/nnet3/chain/build_tree.sh --frame-subsampling-factor 3 \
+      --context-opts "--context-width=2 --central-position=1" \
+      --cmd "$train_cmd" 7000 data/$train_set $lang exp/$ali_dir $treedir
+fi
+
+if [ $stage -le 14 ]; 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)
+  affine_opts="l2-regularize=0.01 dropout-proportion=0.0 dropout-per-dim=true dropout-per-dim-continuous=true"
+  tdnnf_opts="l2-regularize=0.01 dropout-proportion=0.0 bypass-scale=0.66"
+  linear_opts="l2-regularize=0.01 orthonormal-constraint=-1.0"
+  prefinal_opts="l2-regularize=0.01"
+  output_opts="l2-regularize=0.002"
+
+  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-dropout-layer name=tdnn1 $affine_opts dim=1536
+  tdnnf-layer name=tdnnf2 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=1
+  tdnnf-layer name=tdnnf3 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=1
+  tdnnf-layer name=tdnnf4 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=1
+  tdnnf-layer name=tdnnf5 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=0
+  tdnnf-layer name=tdnnf6 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  tdnnf-layer name=tdnnf7 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  tdnnf-layer name=tdnnf8 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  tdnnf-layer name=tdnnf9 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  tdnnf-layer name=tdnnf10 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  tdnnf-layer name=tdnnf11 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  tdnnf-layer name=tdnnf12 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  tdnnf-layer name=tdnnf13 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  tdnnf-layer name=tdnnf14 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  tdnnf-layer name=tdnnf15 $tdnnf_opts dim=1536 bottleneck-dim=160 time-stride=3
+  linear-component name=prefinal-l dim=256 $linear_opts
+
+  prefinal-layer name=prefinal-chain input=prefinal-l $prefinal_opts big-dim=1536 small-dim=256
+  output-layer name=output include-log-softmax=false dim=$num_targets $output_opts
+
+  prefinal-layer name=prefinal-xent input=prefinal-l $prefinal_opts big-dim=1536 small-dim=256
+  output-layer name=output-xent dim=$num_targets learning-rate-factor=$learning_rate_factor $output_opts
+EOF
+  steps/nnet3/xconfig_to_configs.py --xconfig-file $dir/configs/network.xconfig --config-dir $dir/configs/
+fi
+
+if [ $stage -le 15 ]; 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 $train_stage \
+    --cmd "$train_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 0.0 \
+    --chain.apply-deriv-weights false \
+    --chain.lm-opts="--num-extra-lm-states=2000" \
+    --trainer.dropout-schedule $dropout_schedule \
+    --trainer.add-option="--optimization.memory-compression-level=2" \
+    --egs.dir "$common_egs_dir" \
+    --egs.stage $get_egs_stage \
+    --egs.opts "--frames-overlap-per-eg 0 --constrained false" \
+    --egs.chunk-width $frames_per_eg \
+    --trainer.num-chunk-per-minibatch 64 \
+    --trainer.frames-per-iter 1500000 \
+    --trainer.num-epochs $num_epochs \
+    --trainer.optimization.num-jobs-initial 3 \
+    --trainer.optimization.num-jobs-final 16 \
+    --trainer.optimization.initial-effective-lrate 0.00025 \
+    --trainer.optimization.final-effective-lrate 0.000025 \
+    --trainer.max-param-change 2.0 \
+    --cleanup.remove-egs $remove_egs \
+    --feat-dir data/${train_set}_hires \
+    --tree-dir $treedir \
+    --lat-dir exp/tri4_lats_nodup$suffix \
+    --dir $dir  || exit 1;
+
+fi
+
+if [ $stage -le 16 ]; 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_sw1_tg $dir $dir/graph_sw1_tg
+fi
+
+
+graph_dir=$dir/graph_sw1_tg
+iter_opts=
+if [ ! -z $decode_iter ]; then
+  iter_opts=" --iter $decode_iter "
+fi
+if [ $stage -le 17 ]; then
+  rm $dir/.error 2>/dev/null || true
+  for decode_set in train_dev eval2000 $maybe_rt03; do
+      (
+      steps/nnet3/decode.sh --acwt 1.0 --post-decode-acwt 10.0 \
+          --nj $decode_nj --cmd "$decode_cmd" $iter_opts \
+          --online-ivector-dir exp/nnet3/ivectors_${decode_set} \
+          $graph_dir data/${decode_set}_hires \
+          $dir/decode_${decode_set}${decode_iter:+_$decode_iter}_sw1_tg || exit 1;
+      if $has_fisher; then
+          steps/lmrescore_const_arpa.sh --cmd "$decode_cmd" \
+            data/lang_sw1_{tg,fsh_fg} data/${decode_set}_hires \
+            $dir/decode_${decode_set}${decode_iter:+_$decode_iter}_sw1_{tg,fsh_fg} || exit 1;
+      fi
+      ) || touch $dir/.error &
+  done
+  wait
+  if [ -f $dir/.error ]; then
+    echo "$0: something went wrong in decoding"
+    exit 1
+  fi
+fi
+
+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 train_dev eval2000 $maybe_rt03; do
+    (
+      # note: we just give it "$decode_set" as it only uses the wav.scp, the
+      # feature type does not matter.
+
+      steps/online/nnet3/decode.sh --nj $decode_nj --cmd "$decode_cmd" \
+          --acwt 1.0 --post-decode-acwt 10.0 \
+         $graph_dir data/${decode_set}_hires \
+         ${dir}_online/decode_${decode_set}${decode_iter:+_$decode_iter}_sw1_tg || exit 1;
+      if $has_fisher; then
+          steps/lmrescore_const_arpa.sh --cmd "$decode_cmd" \
+            data/lang_sw1_{tg,fsh_fg} data/${decode_set}_hires \
+            ${dir}_online/decode_${decode_set}${decode_iter:+_$decode_iter}_sw1_{tg,fsh_fg} || exit 1;
+      fi
+    ) || touch $dir/.error &
+  done
+  wait
+  if [ -f $dir/.error ]; then
+    echo "$0: something went wrong in decoding"
+    exit 1
+  fi
+fi
+
+
+exit 0;

egs/swbd/s5c/local/chain/run_tdnn_multi_style.sh

@@ -0,0 +1 @@
+multi_style/tuning/run_tdnn_1a.sh

egs/swbd/s5c/local/copy_ali_dir.sh

@@ -0,0 +1,60 @@
+#!/bin/bash
+
+noise_list="reverb1:babble:music:noise"
+max_jobs_run=50
+nj=100
+cmd=queue.pl
+write_binary=true
+
+. ./path.sh
+. utils/parse_options.sh
+
+if [ $# -ne 3 ]; then
+  echo "Usage: $0 <out-data> <src-ali-dir> <out-ali-dir>"
+  exit 1
+fi
+
+data=$1
+src_dir=$2
+dir=$3
+
+mkdir -p $dir
+
+num_jobs=$(cat $src_dir/num_jobs)
+
+rm -f $dir/ali_tmp.*.{ark,scp} 2>/dev/null
+
+# Copy the alignments temporarily
+echo "creating temporary alignments in $dir"
+$cmd --max-jobs-run $max_jobs_run JOB=1:$num_jobs $dir/log/copy_ali_temp.JOB.log \
+  copy-int-vector --binary=$write_binary \
+  "ark:gunzip -c $src_dir/ali.JOB.gz |" \
+  ark,scp:$dir/ali_tmp.JOB.ark,$dir/ali_tmp.JOB.scp || exit 1
+
+# Make copies of utterances for perturbed data
+utt_prefixes=`echo $noise_list | awk -F ":" '{for (i=1; i<=NF; i++) printf "%s- ", $i}'`
+for p in $utt_prefixes; do
+  cat $dir/ali_tmp.*.scp | awk -v p=$p '{print p$0}'
+done | sort -k1,1 > $dir/ali_out.scp.noise
+
+cat $dir/ali_tmp.*.scp | awk '{print $0}' | sort -k1,1 > $dir/ali_out.scp.clean
+
+cat $dir/ali_out.scp.clean $dir/ali_out.scp.noise | sort -k1,1 > $dir/ali_out.scp
+
+utils/split_data.sh ${data} $nj
+
+# Copy and dump the lattices for perturbed data
+echo Creating alignments for augmented data by copying alignments from clean data
+$cmd --max-jobs-run $max_jobs_run JOB=1:$nj $dir/log/copy_out_ali.JOB.log \
+  copy-int-vector --binary=$write_binary \
+  "scp:utils/filter_scp.pl ${data}/split$nj/JOB/utt2spk $dir/ali_out.scp |" \
+  "ark:| gzip -c > $dir/ali.JOB.gz" || exit 1
+
+#rm $dir/ali_out.scp.{prefix,clean} $dir/ali_out.scp
+rm $dir/ali_tmp.*
+
+echo $nj > $dir/num_jobs
+
+for f in cmvn_opts tree splice_opts phones.txt final.mdl splice_opts tree frame_subsampling_factor; do
+  if [ -f $src_dir/$f ]; then cp $src_dir/$f $dir/$f; fi
+done

egs/swbd/s5c/local/copy_lat_dir.sh

@@ -0,0 +1,60 @@
+#!/bin/bash
+
+noise_list="reverb1:babble:music:noise"
+max_jobs_run=50
+nj=100
+cmd=queue.pl
+write_compact=true
+
+. ./path.sh
+. utils/parse_options.sh
+
+if [ $# -ne 3 ]; then
+  echo "Usage: $0 <out-data> <src-ali-dir> <out-ali-dir>"
+  exit 1
+fi
+
+data=$1
+src_dir=$2
+dir=$3
+
+mkdir -p $dir
+
+num_jobs=$(cat $src_dir/num_jobs)
+
+rm -f $dir/lat_tmp.*.{ark,scp} 2>/dev/null
+
+# Copy the alignments temporarily
+echo "creating temporary lattices in $dir"
+$cmd --max-jobs-run $max_jobs_run JOB=1:$num_jobs $dir/log/copy_lat_temp.JOB.log \
+  lattice-copy --write-compact=$write_compact \
+  "ark:gunzip -c $src_dir/lat.JOB.gz |" \
+  ark,scp:$dir/lat_tmp.JOB.ark,$dir/lat_tmp.JOB.scp || exit 1
+
+# Make copies of utterances for perturbed data
+utt_prefixes=`echo $noise_list | awk -F ":" '{for (i=1; i<=NF; i++) printf "%s- ", $i}'`
+for p in $utt_prefixes; do
+  cat $dir/lat_tmp.*.scp | awk -v p=$p '{print p$0}'
+done | sort -k1,1 > $dir/lat_out.scp.noise
+
+cat $dir/lat_tmp.*.scp | awk '{print $0}' | sort -k1,1 > $dir/lat_out.scp.clean
+
+cat $dir/lat_out.scp.clean $dir/lat_out.scp.noise | sort -k1,1 > $dir/lat_out.scp
+
+utils/split_data.sh ${data} $nj
+
+# Copy and dump the lattices for perturbed data
+echo Creating lattices for augmented data by copying lattices from clean data
+$cmd --max-jobs-run $max_jobs_run JOB=1:$nj $dir/log/copy_out_lat.JOB.log \
+  lattice-copy --write-compact=$write_compact \
+  "scp:utils/filter_scp.pl ${data}/split$nj/JOB/utt2spk $dir/lat_out.scp |" \
+  "ark:| gzip -c > $dir/lat.JOB.gz" || exit 1
+
+#rm $dir/lat_out.scp.{noise,clean} $dir/lat_out.scp
+rm $dir/lat_tmp.*
+
+echo $nj > $dir/num_jobs
+
+for f in cmvn_opts splice_opts final.mdl splice_opts tree frame_subsampling_factor; do
+  if [ -f $src_dir/$f ]; then cp $src_dir/$f $dir/$f; fi
+done