From 4ff8e61e4c5dfbb5bc5256614512b3ffcb2d4d76 Mon Sep 17 00:00:00 2001
From: Shinji Watanabe <sw005320@gmail.com>
Date: Thu, 28 Dec 2017 13:04:16 -0500
Subject: [PATCH 1/3] prepare chime4 1ch, 2ch, and 6ch track recipes

---
 egs/chime4/{asr1_1ch => asr1}/RESULTS         |   0
 egs/chime4/{asr1_1ch => asr1}/cmd.sh          |   0
 egs/chime4/{asr1_1ch => asr1}/conf/fbank.conf |   0
 egs/chime4/{asr1_1ch => asr1}/conf/gpu.conf   |   0
 egs/chime4/{asr1_1ch => asr1}/conf/pitch.conf |   0
 .../local/clean_chime4_format_data.sh         |   0
 .../local/clean_wsj0_data_prep.sh             |   0
 .../local/cstr_ndx2flist.pl                   |   0
 .../local/find_noisy_transcripts.pl           |   0
 .../local/find_transcripts.pl                 |   0
 .../{asr1_1ch => asr1}/local/flist2scp.pl     |   0
 .../{asr1_1ch => asr1}/local/make_stft.sh     |   0
 .../local/normalize_transcript.pl             |   0
 .../local/real_enhan_chime4_data_prep.sh      |   0
 .../local/real_noisy_chime4_data_prep.sh      |   0
 .../asr1/local/run_beamform_2ch_track.sh      |  83 +++++++++++++++
 .../asr1/local/run_beamform_6ch_track.sh      | 100 ++++++++++++++++++
 .../local/simu_enhan_chime4_data_prep.sh      |   0
 .../local/simu_noisy_chime4_data_prep.sh      |   0
 egs/chime4/{asr1_1ch => asr1}/path.sh         |   0
 egs/chime4/{asr1_1ch => asr1}/run.sh          |  41 ++++---
 egs/chime4/{asr1_1ch => asr1}/steps           |   0
 egs/chime4/{asr1_1ch => asr1}/utils           |   0
 23 files changed, 209 insertions(+), 15 deletions(-)
 rename egs/chime4/{asr1_1ch => asr1}/RESULTS (100%)
 rename egs/chime4/{asr1_1ch => asr1}/cmd.sh (100%)
 rename egs/chime4/{asr1_1ch => asr1}/conf/fbank.conf (100%)
 rename egs/chime4/{asr1_1ch => asr1}/conf/gpu.conf (100%)
 rename egs/chime4/{asr1_1ch => asr1}/conf/pitch.conf (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/clean_chime4_format_data.sh (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/clean_wsj0_data_prep.sh (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/cstr_ndx2flist.pl (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/find_noisy_transcripts.pl (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/find_transcripts.pl (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/flist2scp.pl (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/make_stft.sh (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/normalize_transcript.pl (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/real_enhan_chime4_data_prep.sh (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/real_noisy_chime4_data_prep.sh (100%)
 create mode 100755 egs/chime4/asr1/local/run_beamform_2ch_track.sh
 create mode 100755 egs/chime4/asr1/local/run_beamform_6ch_track.sh
 rename egs/chime4/{asr1_1ch => asr1}/local/simu_enhan_chime4_data_prep.sh (100%)
 rename egs/chime4/{asr1_1ch => asr1}/local/simu_noisy_chime4_data_prep.sh (100%)
 rename egs/chime4/{asr1_1ch => asr1}/path.sh (100%)
 rename egs/chime4/{asr1_1ch => asr1}/run.sh (79%)
 rename egs/chime4/{asr1_1ch => asr1}/steps (100%)
 rename egs/chime4/{asr1_1ch => asr1}/utils (100%)

diff --git a/egs/chime4/asr1_1ch/RESULTS b/egs/chime4/asr1/RESULTS
similarity index 100%
rename from egs/chime4/asr1_1ch/RESULTS
rename to egs/chime4/asr1/RESULTS
diff --git a/egs/chime4/asr1_1ch/cmd.sh b/egs/chime4/asr1/cmd.sh
similarity index 100%
rename from egs/chime4/asr1_1ch/cmd.sh
rename to egs/chime4/asr1/cmd.sh
diff --git a/egs/chime4/asr1_1ch/conf/fbank.conf b/egs/chime4/asr1/conf/fbank.conf
similarity index 100%
rename from egs/chime4/asr1_1ch/conf/fbank.conf
rename to egs/chime4/asr1/conf/fbank.conf
diff --git a/egs/chime4/asr1_1ch/conf/gpu.conf b/egs/chime4/asr1/conf/gpu.conf
similarity index 100%
rename from egs/chime4/asr1_1ch/conf/gpu.conf
rename to egs/chime4/asr1/conf/gpu.conf
diff --git a/egs/chime4/asr1_1ch/conf/pitch.conf b/egs/chime4/asr1/conf/pitch.conf
similarity index 100%
rename from egs/chime4/asr1_1ch/conf/pitch.conf
rename to egs/chime4/asr1/conf/pitch.conf
diff --git a/egs/chime4/asr1_1ch/local/clean_chime4_format_data.sh b/egs/chime4/asr1/local/clean_chime4_format_data.sh
similarity index 100%
rename from egs/chime4/asr1_1ch/local/clean_chime4_format_data.sh
rename to egs/chime4/asr1/local/clean_chime4_format_data.sh
diff --git a/egs/chime4/asr1_1ch/local/clean_wsj0_data_prep.sh b/egs/chime4/asr1/local/clean_wsj0_data_prep.sh
similarity index 100%
rename from egs/chime4/asr1_1ch/local/clean_wsj0_data_prep.sh
rename to egs/chime4/asr1/local/clean_wsj0_data_prep.sh
diff --git a/egs/chime4/asr1_1ch/local/cstr_ndx2flist.pl b/egs/chime4/asr1/local/cstr_ndx2flist.pl
similarity index 100%
rename from egs/chime4/asr1_1ch/local/cstr_ndx2flist.pl
rename to egs/chime4/asr1/local/cstr_ndx2flist.pl
diff --git a/egs/chime4/asr1_1ch/local/find_noisy_transcripts.pl b/egs/chime4/asr1/local/find_noisy_transcripts.pl
similarity index 100%
rename from egs/chime4/asr1_1ch/local/find_noisy_transcripts.pl
rename to egs/chime4/asr1/local/find_noisy_transcripts.pl
diff --git a/egs/chime4/asr1_1ch/local/find_transcripts.pl b/egs/chime4/asr1/local/find_transcripts.pl
similarity index 100%
rename from egs/chime4/asr1_1ch/local/find_transcripts.pl
rename to egs/chime4/asr1/local/find_transcripts.pl
diff --git a/egs/chime4/asr1_1ch/local/flist2scp.pl b/egs/chime4/asr1/local/flist2scp.pl
similarity index 100%
rename from egs/chime4/asr1_1ch/local/flist2scp.pl
rename to egs/chime4/asr1/local/flist2scp.pl
diff --git a/egs/chime4/asr1_1ch/local/make_stft.sh b/egs/chime4/asr1/local/make_stft.sh
similarity index 100%
rename from egs/chime4/asr1_1ch/local/make_stft.sh
rename to egs/chime4/asr1/local/make_stft.sh
diff --git a/egs/chime4/asr1_1ch/local/normalize_transcript.pl b/egs/chime4/asr1/local/normalize_transcript.pl
similarity index 100%
rename from egs/chime4/asr1_1ch/local/normalize_transcript.pl
rename to egs/chime4/asr1/local/normalize_transcript.pl
diff --git a/egs/chime4/asr1_1ch/local/real_enhan_chime4_data_prep.sh b/egs/chime4/asr1/local/real_enhan_chime4_data_prep.sh
similarity index 100%
rename from egs/chime4/asr1_1ch/local/real_enhan_chime4_data_prep.sh
rename to egs/chime4/asr1/local/real_enhan_chime4_data_prep.sh
diff --git a/egs/chime4/asr1_1ch/local/real_noisy_chime4_data_prep.sh b/egs/chime4/asr1/local/real_noisy_chime4_data_prep.sh
similarity index 100%
rename from egs/chime4/asr1_1ch/local/real_noisy_chime4_data_prep.sh
rename to egs/chime4/asr1/local/real_noisy_chime4_data_prep.sh
diff --git a/egs/chime4/asr1/local/run_beamform_2ch_track.sh b/egs/chime4/asr1/local/run_beamform_2ch_track.sh
new file mode 100755
index 00000000000..a974adbcad6
--- /dev/null
+++ b/egs/chime4/asr1/local/run_beamform_2ch_track.sh
@@ -0,0 +1,83 @@
+#!/bin/bash
+
+# Copyright 2015, Mitsubishi Electric Research Laboratories, MERL (Author: Shinji Watanabe)
+
+. ./cmd.sh
+. ./path.sh
+
+# Config:
+nj=10
+cmd=run.pl
+
+. utils/parse_options.sh || exit 1;
+
+if [ $# != 2 ]; then
+   echo "Wrong #arguments ($#, expected 3)"
+   echo "Usage: local/run_beamform_2ch_track.sh [options] <wav-in-dir> <wav-out-dir>"
+   echo "main options (for others, see top of script file)"
+   echo "  --nj <nj>                                # number of parallel jobs"
+   echo "  --cmd <cmd>                              # Command to run in parallel with"
+   exit 1;
+fi
+
+sdir=$1
+odir=$2
+
+wdir=data/beamforming_2ch_track
+
+if [ -z $BEAMFORMIT ] ; then
+  export BEAMFORMIT=$KALDI_ROOT/tools/extras/BeamformIt
+fi
+export PATH=${PATH}:$BEAMFORMIT
+! hash BeamformIt && echo "Missing BeamformIt, run 'cd ../../../tools/kaldi/tools; extras/install_beamformit.sh; cd -;'" && exit 1
+
+# Set bash to 'debug' mode, it will exit on :
+# -e 'error', -u 'undefined variable', -o ... 'error in pipeline', -x 'print commands',
+set -e
+set -u
+set -o pipefail
+
+mkdir -p $odir
+mkdir -p $wdir/log
+
+allwavs=`find $sdir/ | grep "\.wav" | tr ' ' '\n' | awk -F '/' '{print $(NF-1)"/"$NF}'`
+
+# wavfiles.list can be used as the name of the output files
+output_wavfiles=$wdir/wavfiles.list
+echo $allwavs | tr ' ' '\n' | awk -F '.' '{print $1}' | sort | uniq > $output_wavfiles
+
+# channel list
+input_arrays=$wdir/channels
+echo $allwavs | tr ' ' '\n' | sort | awk 'NR%2==1' > $wdir/channels.1st
+echo $allwavs | tr ' ' '\n' | sort | awk 'NR%2==0' > $wdir/channels.2nd
+paste -d" " $output_wavfiles $wdir/channels.1st $wdir/channels.2nd > $input_arrays
+
+# split the list for parallel processing
+split_wavfiles=""
+for n in `seq $nj`; do
+  split_wavfiles="$split_wavfiles $output_wavfiles.$n"
+done
+utils/split_scp.pl $output_wavfiles $split_wavfiles || exit 1;
+
+echo -e "Beamforming\n"
+# making a shell script for each job
+for n in `seq $nj`; do
+cat << EOF > $wdir/log/beamform.$n.sh
+while read line; do
+  $BEAMFORMIT/BeamformIt -s \$line -c $input_arrays \
+    --config_file `pwd`/conf/chime4.cfg \
+    --source_dir $sdir \
+    --result_dir $odir
+done < $output_wavfiles.$n
+EOF
+done
+# making a subdirectory for the output wav files
+for x in `awk -F '/' '{print $1}' $output_wavfiles | sort | uniq`; do
+  mkdir -p $odir/$x
+done
+
+chmod a+x $wdir/log/beamform.*.sh
+$cmd JOB=1:$nj $wdir/log/beamform.JOB.log \
+  $wdir/log/beamform.JOB.sh
+
+echo "`basename $0` Done."
diff --git a/egs/chime4/asr1/local/run_beamform_6ch_track.sh b/egs/chime4/asr1/local/run_beamform_6ch_track.sh
new file mode 100755
index 00000000000..5ce5e423391
--- /dev/null
+++ b/egs/chime4/asr1/local/run_beamform_6ch_track.sh
@@ -0,0 +1,100 @@
+#!/bin/bash
+
+# Copyright 2015, Mitsubishi Electric Research Laboratories, MERL (Author: Shinji Watanabe)
+
+. ./cmd.sh
+. ./path.sh
+
+# Config:
+nj=10
+cmd=run.pl
+bmf="1 3 4 5 6"
+eval_flag=true # make it true when the evaluation data are released
+
+. utils/parse_options.sh || exit 1;
+
+if [ $# != 2 ]; then
+   echo "Wrong #arguments ($#, expected 2)"
+   echo "Usage: local/run_beamform_6ch_track.sh [options] <wav-in-dir> <wav-out-dir>"
+   echo "main options (for others, see top of script file)"
+   echo "  --nj <nj>                                # number of parallel jobs"
+   echo "  --cmd <cmd>                              # Command to run in parallel with"
+   echo "  --bmf \"1 3 4 5 6\"                      # microphones used for beamforming (2th mic is omitted in default)"
+   exit 1;
+fi
+
+sdir=$1
+odir=$2
+wdir=data/beamforming_`echo $bmf | tr ' ' '_'`
+
+if [ -z $BEAMFORMIT ] ; then
+  export BEAMFORMIT=$KALDI_ROOT/tools/extras/BeamformIt
+fi
+export PATH=${PATH}:$BEAMFORMIT
+! hash BeamformIt && echo "Missing BeamformIt, run 'cd ../../../tools/kaldi/tools; extras/install_beamformit.sh; cd -;'" && exit 1
+
+# Set bash to 'debug' mode, it will exit on :
+# -e 'error', -u 'undefined variable', -o ... 'error in pipeline', -x 'print commands',
+set -e
+set -u
+set -o pipefail
+
+mkdir -p $odir
+mkdir -p $wdir/log
+
+echo "Will use the following channels: $bmf"
+# number of channels
+numch=`echo $bmf | tr ' ' '\n' | wc -l`
+echo "the number of channels: $numch"
+
+# wavfiles.list can be used as the name of the output files
+# we only process dev and eval waves
+output_wavfiles=$wdir/wavfiles.list
+if $eval_flag; then
+  find $sdir/{dt,et}*{simu,real}/ | grep CH1.wav \
+    | awk -F '/' '{print $(NF-1) "/" $NF}' | sed -e "s/\.CH1\.wav//" | sort > $output_wavfiles
+else
+  find $sdir/dt*{simu,real}/ | grep CH1.wav \
+    | awk -F '/' '{print $(NF-1) "/" $NF}' | sed -e "s/\.CH1\.wav//" | sort > $output_wavfiles
+fi
+
+# this is an input file list of the microphones
+# format: 1st_wav 2nd_wav ... nth_wav
+input_arrays=$wdir/channels_$numch
+for x in `cat $output_wavfiles`; do
+  echo -n "$x"
+  for ch in $bmf; do
+    echo -n " $x.CH$ch.wav"
+  done
+  echo ""
+done > $input_arrays
+
+# split the list for parallel processing
+split_wavfiles=""
+for n in `seq $nj`; do
+  split_wavfiles="$split_wavfiles $output_wavfiles.$n"
+done
+utils/split_scp.pl $output_wavfiles $split_wavfiles || exit 1;
+
+echo -e "Beamforming\n"
+# making a shell script for each job
+for n in `seq $nj`; do
+cat << EOF > $wdir/log/beamform.$n.sh
+while read line; do
+  $BEAMFORMIT/BeamformIt -s \$line -c $input_arrays \
+    --config_file `pwd`/conf/chime4.cfg \
+    --source_dir $sdir \
+    --result_dir $odir
+done < $output_wavfiles.$n
+EOF
+done
+# making a subdirectory for the output wav files
+for x in `awk -F '/' '{print $1}' $output_wavfiles | sort | uniq`; do
+  mkdir -p $odir/$x
+done
+
+chmod a+x $wdir/log/beamform.*.sh
+$cmd JOB=1:$nj $wdir/log/beamform.JOB.log \
+  $wdir/log/beamform.JOB.sh
+
+echo "`basename $0` Done."
diff --git a/egs/chime4/asr1_1ch/local/simu_enhan_chime4_data_prep.sh b/egs/chime4/asr1/local/simu_enhan_chime4_data_prep.sh
similarity index 100%
rename from egs/chime4/asr1_1ch/local/simu_enhan_chime4_data_prep.sh
rename to egs/chime4/asr1/local/simu_enhan_chime4_data_prep.sh
diff --git a/egs/chime4/asr1_1ch/local/simu_noisy_chime4_data_prep.sh b/egs/chime4/asr1/local/simu_noisy_chime4_data_prep.sh
similarity index 100%
rename from egs/chime4/asr1_1ch/local/simu_noisy_chime4_data_prep.sh
rename to egs/chime4/asr1/local/simu_noisy_chime4_data_prep.sh
diff --git a/egs/chime4/asr1_1ch/path.sh b/egs/chime4/asr1/path.sh
similarity index 100%
rename from egs/chime4/asr1_1ch/path.sh
rename to egs/chime4/asr1/path.sh
diff --git a/egs/chime4/asr1_1ch/run.sh b/egs/chime4/asr1/run.sh
similarity index 79%
rename from egs/chime4/asr1_1ch/run.sh
rename to egs/chime4/asr1/run.sh
index b9e8b6ad6e9..ab7b0437f13 100755
--- a/egs/chime4/asr1_1ch/run.sh
+++ b/egs/chime4/asr1/run.sh
@@ -54,8 +54,6 @@ recog_model=acc.best # set a model to be used for decoding: 'acc.best' or 'loss.
 
 # data
 chime4_data=/export/corpora4/CHiME4/CHiME3 # JHU setup
-enhan=isolated_1ch_track
-enhan_data=$chime4_data/data/audio/16kHz/$enhan
 
 # exp tag
 tag="" # tag for managing experiments.
@@ -72,21 +70,35 @@ set -u
 set -o pipefail
 
 train_set=tr05_multi_noisy
-train_dev=dt05_multi_${enhan}
-recog_set="dt05_multi_${enhan} et05_multi_${enhan}"
+train_dev=dt05_multi_isolated_1ch_track
+recog_set="\
+dt05_real_isolated_1ch_track dt05_simu_isolated_1ch_track et05_real_isolated_1ch_track et05_simu_isolated_1ch_track \
+dt05_real_beamformit_2mics dt05_simu_beamformit_2mics et05_real_beamformit_2mics et05_simu_beamformit_2mics \
+dt05_real_beamformit_5mics dt05_simu_beamformit_5mics et05_real_beamformit_5mics et05_simu_beamformit_5mics \
+"
 
 if [ ${stage} -le 0 ]; then
-    ### Task dependent. You have to make data the following preparation part by yourself.
+    ### Task dependent. You have to make the following data preparation part by yourself.
     ### But you can utilize Kaldi recipes in most cases
-     echo "stage 0: Data preparation"
+    echo "stage 0: Data preparation"
     wsj0_data=${chime4_data}/data/WSJ0
     local/clean_wsj0_data_prep.sh ${wsj0_data}
     local/clean_chime4_format_data.sh
+    echo "beamforming for multichannel cases"
+    local/run_beamform_2ch_track.sh --cmd "${train_cmd}" --nj 20 ${chime4_data}/data/audio/16kHz/isolated_2ch_track enhan/beamformit_2mics
+    local/run_beamform_6ch_track.sh --cmd "${train_cmd}" --nj 20 ${chime4_data}/data/audio/16kHz/isolated_6ch_track enhan/beamformit_5mics
     echo "prepartion for chime4 data"
     local/real_noisy_chime4_data_prep.sh ${chime4_data}
     local/simu_noisy_chime4_data_prep.sh ${chime4_data}
-    local/real_enhan_chime4_data_prep.sh ${enhan} ${enhan_data}
-    local/simu_enhan_chime4_data_prep.sh ${enhan} ${enhan_data}
+    echo "test data for 1ch track"
+    local/real_enhan_chime4_data_prep.sh isolated_1ch_track ${chime4_data}/data/audio/16kHz/isolated_1ch_track
+    local/simu_enhan_chime4_data_prep.sh isolated_1ch_track ${chime4_data}/data/audio/16kHz/isolated_1ch_track
+    echo "test data for 2ch track"
+    local/real_enhan_chime4_data_prep.sh beamformit_2mics enhan/beamformit_2mics
+    local/simu_enhan_chime4_data_prep.sh beamformit_2mics enhan/beamformit_2mics
+    echo "test data for 6ch track"
+    local/real_enhan_chime4_data_prep.sh beamformit_5mics enhan/beamformit_5mics
+    local/simu_enhan_chime4_data_prep.sh beamformit_5mics enhan/beamformit_5mics
 fi
 
 feat_tr_dir=${dumpdir}/${train_set}/delta${do_delta}; mkdir -p ${feat_tr_dir}
@@ -94,10 +106,10 @@ feat_dt_dir=${dumpdir}/${train_dev}/delta${do_delta}; mkdir -p ${feat_dt_dir}
 if [ ${stage} -le 1 ]; then
     ### Task dependent. You have to design training and dev sets by yourself.
     ### But you can utilize Kaldi recipes in most cases
-     echo "stage 1: Feature Generation"
+    echo "stage 1: Feature Generation"
     # Generate the fbank features; by default 80-dimensional fbanks with pitch on each frame
     fbankdir=fbank
-    tasks="tr05_real_noisy tr05_simu_noisy dt05_real_${enhan} dt05_simu_${enhan} et05_real_${enhan} et05_simu_${enhan}"
+    tasks="tr05_real_noisy tr05_simu_noisy ${recog_set}"
     for x in ${tasks}; do
         utils/copy_data_dir.sh data/${x} data-fbank/${x}
         utils/copy_data_dir.sh data/${x} data-stft/${x}
@@ -105,9 +117,8 @@ if [ ${stage} -le 1 ]; then
     done
 
     echo "combine real and simulation data"
-    utils/combine_data.sh data-fbank/tr05_multi_noisy data-fbank/tr05_simu_noisy data-fbank/tr05_real_noisy
-    utils/combine_data.sh data-fbank/dt05_multi_${enhan} data-fbank/dt05_simu_${enhan} data-fbank/dt05_real_${enhan}
-    utils/combine_data.sh data-fbank/et05_multi_${enhan} data-fbank/et05_simu_${enhan} data-fbank/et05_real_${enhan}
+    utils/combine_data.sh data-fbank/${train_set} data-fbank/tr05_simu_noisy data-fbank/tr05_real_noisy
+    utils/combine_data.sh data-fbank/${train_dev} data-fbank/dt05_simu_isolated_1ch_track data-fbank/dt05_real_isolated_1ch_track
 
     # compute global CMVN
     compute-cmvn-stats scp:data-fbank/${train_set}/feats.scp data-fbank/${train_set}/cmvn.ark
@@ -115,12 +126,12 @@ if [ ${stage} -le 1 ]; then
     # dump features for training
     if [[ $(hostname -f) == *.clsp.jhu.edu ]] && [ ! -d ${feat_tr_dir}/storage ]; then
     utils/create_split_dir.pl \
-        /export/b{10,11,12,13}/${USER}/espnet-data/egs/voxforge/asr1/dump/${train_set}/delta${do_delta}/storage \
+        /export/b{14,15,16,17}/${USER}/espnet-data/egs/chime4/asr1/dump/${train_set}/delta${do_delta}/storage \
         ${feat_tr_dir}/storage
     fi
     if [[ $(hostname -f) == *.clsp.jhu.edu ]] && [ ! -d ${feat_dt_dir}/storage ]; then
     utils/create_split_dir.pl \
-        /export/b{10,11,12,13}/${USER}/espnet-data/egs/voxforge/asr1/dump/${train_dev}/delta${do_delta}/storage \
+        /export/b{14,15,16,17}/${USER}/espnet-data/egs/chime4/asr1/dump/${train_dev}/delta${do_delta}/storage \
         ${feat_dt_dir}/storage
     fi
     dump.sh --cmd "$train_cmd" --nj 32 --do_delta $do_delta \
diff --git a/egs/chime4/asr1_1ch/steps b/egs/chime4/asr1/steps
similarity index 100%
rename from egs/chime4/asr1_1ch/steps
rename to egs/chime4/asr1/steps
diff --git a/egs/chime4/asr1_1ch/utils b/egs/chime4/asr1/utils
similarity index 100%
rename from egs/chime4/asr1_1ch/utils
rename to egs/chime4/asr1/utils

From f8a5ad42be214bed266fbf1aa83b9bc32216c081 Mon Sep 17 00:00:00 2001
From: Shinji Watanabe <sw005320@gmail.com>
Date: Thu, 28 Dec 2017 14:42:16 -0500
Subject: [PATCH 2/3] fix bugs

---
 egs/chime4/asr1/conf/chime4.cfg               | 50 +++++++++++++++++++
 egs/chime4/asr1/local/clean_wsj0_data_prep.sh |  3 +-
 .../asr1/local/run_beamform_2ch_track.sh      |  2 +-
 .../asr1/local/run_beamform_6ch_track.sh      |  2 +-
 egs/chime4/asr1/run.sh                        |  6 ++-
 5 files changed, 57 insertions(+), 6 deletions(-)
 create mode 100755 egs/chime4/asr1/conf/chime4.cfg

diff --git a/egs/chime4/asr1/conf/chime4.cfg b/egs/chime4/asr1/conf/chime4.cfg
new file mode 100755
index 00000000000..70fdd858651
--- /dev/null
+++ b/egs/chime4/asr1/conf/chime4.cfg
@@ -0,0 +1,50 @@
+#BeamformIt sample configuration file for AMI data (http://groups.inf.ed.ac.uk/ami/download/)
+
+# scrolling size to compute the delays
+scroll_size = 250
+
+# cross correlation computation window size
+window_size = 500
+
+#amount of maximum points for the xcorrelation taken into account
+nbest_amount = 4
+
+#flag wether to apply an automatic noise thresholding 
+do_noise_threshold = 1
+
+#Percentage of frames with lower xcorr taken as noisy
+noise_percent = 10
+
+######## acoustic modelling parameters
+
+#transition probabilities weight for multichannel decoding
+trans_weight_multi = 25
+trans_weight_nbest = 25
+
+###
+
+#flag wether to print the feaures after setting them, or not
+print_features = 1
+
+#flag wether to use the bad frames in the sum process
+do_avoid_bad_frames = 1
+
+#flag to use the best channel (SNR) as a reference
+#defined from command line
+do_compute_reference = 1
+
+#flag wether to use a uem file or not(process all the file)
+do_use_uem_file = 0
+
+#flag wether to use an adaptative weights scheme or fixed weights
+do_adapt_weights = 1
+
+#flag wether to output the sph files or just run the system to create the auxiliary files
+do_write_sph_files = 1
+
+####directories where to store/retrieve info####
+#channels_file = ./cfg-files/channels
+
+#show needs to be passed as argument normally, here a default one is given just in case
+#show_id = Ttmp
+
diff --git a/egs/chime4/asr1/local/clean_wsj0_data_prep.sh b/egs/chime4/asr1/local/clean_wsj0_data_prep.sh
index d1d15568de1..170b77612d4 100755
--- a/egs/chime4/asr1/local/clean_wsj0_data_prep.sh
+++ b/egs/chime4/asr1/local/clean_wsj0_data_prep.sh
@@ -25,8 +25,7 @@ dir=`pwd`/data/local/data
 local=`pwd`/local
 utils=`pwd`/utils
 
-. ./path.sh # Needed for KALDI_ROOT
-sph2pipe=${KALDI_ROOT}/tools/sph2pipe_v2.5/sph2pipe
+sph2pipe=../../../tools/kaldi/tools/sph2pipe_v2.5/sph2pipe
 if [ ! -x ${sph2pipe} ]; then
   echo "Could not find (or execute) the sph2pipe program at ${sph2pipe}";
   exit 1;
diff --git a/egs/chime4/asr1/local/run_beamform_2ch_track.sh b/egs/chime4/asr1/local/run_beamform_2ch_track.sh
index a974adbcad6..2d03178a6e0 100755
--- a/egs/chime4/asr1/local/run_beamform_2ch_track.sh
+++ b/egs/chime4/asr1/local/run_beamform_2ch_track.sh
@@ -26,7 +26,7 @@ odir=$2
 wdir=data/beamforming_2ch_track
 
 if [ -z $BEAMFORMIT ] ; then
-  export BEAMFORMIT=$KALDI_ROOT/tools/extras/BeamformIt
+  export BEAMFORMIT=$KALDI_ROOT/tools/BeamformIt
 fi
 export PATH=${PATH}:$BEAMFORMIT
 ! hash BeamformIt && echo "Missing BeamformIt, run 'cd ../../../tools/kaldi/tools; extras/install_beamformit.sh; cd -;'" && exit 1
diff --git a/egs/chime4/asr1/local/run_beamform_6ch_track.sh b/egs/chime4/asr1/local/run_beamform_6ch_track.sh
index 5ce5e423391..6a6f398ca6b 100755
--- a/egs/chime4/asr1/local/run_beamform_6ch_track.sh
+++ b/egs/chime4/asr1/local/run_beamform_6ch_track.sh
@@ -28,7 +28,7 @@ odir=$2
 wdir=data/beamforming_`echo $bmf | tr ' ' '_'`
 
 if [ -z $BEAMFORMIT ] ; then
-  export BEAMFORMIT=$KALDI_ROOT/tools/extras/BeamformIt
+  export BEAMFORMIT=$KALDI_ROOT/tools/BeamformIt
 fi
 export PATH=${PATH}:$BEAMFORMIT
 ! hash BeamformIt && echo "Missing BeamformIt, run 'cd ../../../tools/kaldi/tools; extras/install_beamformit.sh; cd -;'" && exit 1
diff --git a/egs/chime4/asr1/run.sh b/egs/chime4/asr1/run.sh
index ab7b0437f13..bad4a461d3a 100755
--- a/egs/chime4/asr1/run.sh
+++ b/egs/chime4/asr1/run.sh
@@ -85,8 +85,10 @@ if [ ${stage} -le 0 ]; then
     local/clean_wsj0_data_prep.sh ${wsj0_data}
     local/clean_chime4_format_data.sh
     echo "beamforming for multichannel cases"
-    local/run_beamform_2ch_track.sh --cmd "${train_cmd}" --nj 20 ${chime4_data}/data/audio/16kHz/isolated_2ch_track enhan/beamformit_2mics
-    local/run_beamform_6ch_track.sh --cmd "${train_cmd}" --nj 20 ${chime4_data}/data/audio/16kHz/isolated_6ch_track enhan/beamformit_5mics
+    local/run_beamform_2ch_track.sh --cmd "${train_cmd}" --nj 20 \
+        ${chime4_data}/data/audio/16kHz/isolated_2ch_track enhan/beamformit_2mics
+    local/run_beamform_6ch_track.sh --cmd "${train_cmd}" --nj 20 \
+        ${chime4_data}/data/audio/16kHz/isolated_6ch_track enhan/beamformit_5mics
     echo "prepartion for chime4 data"
     local/real_noisy_chime4_data_prep.sh ${chime4_data}
     local/simu_noisy_chime4_data_prep.sh ${chime4_data}

From 186ab03272368c39a4035d229c02073f1f1eba0d Mon Sep 17 00:00:00 2001
From: Shinji Watanabe <sw005320@gmail.com>
Date: Thu, 28 Dec 2017 18:00:37 -0500
Subject: [PATCH 3/3] fix chime4 recipe

---
 egs/chime4/asr1/local/real_enhan_chime4_data_prep.sh | 8 --------
 egs/chime4/asr1/local/real_noisy_chime4_data_prep.sh | 2 --
 egs/chime4/asr1/local/simu_enhan_chime4_data_prep.sh | 8 --------
 egs/chime4/asr1/local/simu_noisy_chime4_data_prep.sh | 2 --
 egs/chime4/asr1/run.sh                               | 8 ++++----
 5 files changed, 4 insertions(+), 24 deletions(-)

diff --git a/egs/chime4/asr1/local/real_enhan_chime4_data_prep.sh b/egs/chime4/asr1/local/real_enhan_chime4_data_prep.sh
index 7d4f9c892a8..cb25370f1a5 100755
--- a/egs/chime4/asr1/local/real_enhan_chime4_data_prep.sh
+++ b/egs/chime4/asr1/local/real_enhan_chime4_data_prep.sh
@@ -33,14 +33,6 @@ local=`pwd`/local
 utils=`pwd`/utils
 odir=`pwd`/data
 
-. ./path.sh # Needed for KALDI_ROOT
-export PATH=$PATH:$KALDI_ROOT/tools/irstlm/bin
-sph2pipe=$KALDI_ROOT/tools/sph2pipe_v2.5/sph2pipe
-if [ ! -x $sph2pipe ]; then
-  echo "Could not find (or execute) the sph2pipe program at $sph2pipe";
-  exit 1;
-fi
-
 if $eval_flag; then
 list_set="tr05_real_$enhan dt05_real_$enhan et05_real_$enhan"
 else
diff --git a/egs/chime4/asr1/local/real_noisy_chime4_data_prep.sh b/egs/chime4/asr1/local/real_noisy_chime4_data_prep.sh
index 6efe151949c..98a8a22a556 100755
--- a/egs/chime4/asr1/local/real_noisy_chime4_data_prep.sh
+++ b/egs/chime4/asr1/local/real_noisy_chime4_data_prep.sh
@@ -37,8 +37,6 @@ dir=`pwd`/data/local/data
 local=`pwd`/local
 utils=`pwd`/utils
 
-. ./path.sh # Needed for KALDI_ROOT
-
 if ${eval_flag}; then
 list_set="tr05_real_noisy dt05_real_noisy et05_real_noisy"
 else
diff --git a/egs/chime4/asr1/local/simu_enhan_chime4_data_prep.sh b/egs/chime4/asr1/local/simu_enhan_chime4_data_prep.sh
index d6419fa90b9..027dd6630e2 100755
--- a/egs/chime4/asr1/local/simu_enhan_chime4_data_prep.sh
+++ b/egs/chime4/asr1/local/simu_enhan_chime4_data_prep.sh
@@ -33,14 +33,6 @@ local=`pwd`/local
 utils=`pwd`/utils
 odir=`pwd`/data
 
-. ./path.sh # Needed for KALDI_ROOT
-export PATH=$PATH:$KALDI_ROOT/tools/irstlm/bin
-sph2pipe=$KALDI_ROOT/tools/sph2pipe_v2.5/sph2pipe
-if [ ! -x $sph2pipe ]; then
-  echo "Could not find (or execute) the sph2pipe program at $sph2pipe";
-  exit 1;
-fi
-
 if $eval_flag; then
 list_set="tr05_simu_$enhan dt05_simu_$enhan et05_simu_$enhan"
 else
diff --git a/egs/chime4/asr1/local/simu_noisy_chime4_data_prep.sh b/egs/chime4/asr1/local/simu_noisy_chime4_data_prep.sh
index 28c754fe3b6..01db3ffab8b 100755
--- a/egs/chime4/asr1/local/simu_noisy_chime4_data_prep.sh
+++ b/egs/chime4/asr1/local/simu_noisy_chime4_data_prep.sh
@@ -37,8 +37,6 @@ dir=`pwd`/data/local/data
 local=`pwd`/local
 utils=`pwd`/utils
 
-. ./path.sh # Needed for KALDI_ROOT
-
 if ${eval_flag}; then
 list_set="tr05_simu_noisy dt05_simu_noisy et05_simu_noisy"
 else
diff --git a/egs/chime4/asr1/run.sh b/egs/chime4/asr1/run.sh
index bad4a461d3a..7cd0ee5e7a5 100755
--- a/egs/chime4/asr1/run.sh
+++ b/egs/chime4/asr1/run.sh
@@ -96,11 +96,11 @@ if [ ${stage} -le 0 ]; then
     local/real_enhan_chime4_data_prep.sh isolated_1ch_track ${chime4_data}/data/audio/16kHz/isolated_1ch_track
     local/simu_enhan_chime4_data_prep.sh isolated_1ch_track ${chime4_data}/data/audio/16kHz/isolated_1ch_track
     echo "test data for 2ch track"
-    local/real_enhan_chime4_data_prep.sh beamformit_2mics enhan/beamformit_2mics
-    local/simu_enhan_chime4_data_prep.sh beamformit_2mics enhan/beamformit_2mics
+    local/real_enhan_chime4_data_prep.sh beamformit_2mics ${PWD}/enhan/beamformit_2mics
+    local/simu_enhan_chime4_data_prep.sh beamformit_2mics ${PWD}/enhan/beamformit_2mics
     echo "test data for 6ch track"
-    local/real_enhan_chime4_data_prep.sh beamformit_5mics enhan/beamformit_5mics
-    local/simu_enhan_chime4_data_prep.sh beamformit_5mics enhan/beamformit_5mics
+    local/real_enhan_chime4_data_prep.sh beamformit_5mics ${PWD}/enhan/beamformit_5mics
+    local/simu_enhan_chime4_data_prep.sh beamformit_5mics ${PWD}/enhan/beamformit_5mics
 fi
 
 feat_tr_dir=${dumpdir}/${train_set}/delta${do_delta}; mkdir -p ${feat_tr_dir}