tomwalters@306: #!/bin/bash
tomwalters@306: #
tomwalters@306: # Run the syllables database 'spider pattern' through AIM-C
tomwalters@306: #
tomwalters@306: # Copyright 2009-2010 University of Cambridge
tomwalters@306: # Author: Thomas Walters <tom@acousticscale.org>
tomwalters@306: 
tomwalters@306: # Source directory for all the sound files
tomwalters@312: #SOUND_SOURCE="/Users/Tom/Documents/Work/PhD/HTk-AIM/Sounds/"
tomwalters@314: #SOUND_SOURCE="/media/sounds-database/cnbh-sounds/"
tomwalters@314: SOUND_SOURCE="/mnt/sounds/snr_+0dB/"
tomwalters@306: 
tomwalters@306: # Location of the AIMCopy binary if not in the path
tomwalters@312: AIMCOPY_PREFIX="../aimc-read-only/build/posix-release/"
tomwalters@306: 
tomwalters@314: MACHINE_CORES=8
tomwalters@306: 
tomwalters@316: # Set to true / 1 to enable MFCC features rather than AIM features
tomwalters@316: # (leave blank for AIM features)
tomwalters@316: MFCC_FEATURES=
tomwalters@316: 
tomwalters@306: # Names of various internal files and directories. 
tomwalters@306: # Rename here if you don't like them for some reason.
tomwalters@306: SYLLIST=syls
tomwalters@306: TRAIN_SPEAKERS=train_speakers
tomwalters@306: TEST_SPEAKERS=test_speakers
tomwalters@306: WORK_PREFIX=snr
tomwalters@306: TRAIN_LIST=train.list
tomwalters@306: TEST_LIST=test.list
tomwalters@306: COMBINED_LIST=combined.list
tomwalters@306: FEATURES_DIR=features
tomwalters@306: AIMCOPY_CONFIG=aimcopy.cfg
tomwalters@316: HCOPY_CONFIG=hcopy.cfg
tomwalters@306: AIMCOPY_LOG_TRAIN=aimcopy_train.log
tomwalters@306: AIMCOPY_LOG_TEST=aimcopy_test.log
tomwalters@316: HTK_PREFIX=""
tomwalters@306: 
tomwalters@306: 
tomwalters@306: # The vowels and consonants that make up the CNBH database
tomwalters@306: VOWELS="a e i o u"
tomwalters@306: CONSONANTS="b d f g h k l m n p r s t v w x y z"
tomwalters@306: SILENCE="sil"
tomwalters@306: 
tomwalters@306: WORK=${WORK_PREFIX}`echo $1 | tr -d ' '`
tomwalters@306: mkdir -p $WORK
tomwalters@306: 
tomwalters@306: # Make a copy of this script in the experimental directory
tomwalters@306: cp -p $0  $WORK
tomwalters@306: 
tomwalters@306: 
tomwalters@306: echo "Generating syllable list..."
tomwalters@306: 
tomwalters@306: for v in $VOWELS; do
tomwalters@308:   echo $v$v >> $WORK/$SYLLIST.tmp
tomwalters@306:   for c in $CONSONANTS; do
tomwalters@308:     echo $v$c >> $WORK/$SYLLIST.tmp
tomwalters@307:     echo $c$v >> $WORK/$SYLLIST.tmp
tomwalters@306:   done
tomwalters@306: done
tomwalters@306: 
tomwalters@306: # Sort the syllable list and delete the 
tomwalters@306: # temporary, unsorted version
tomwalters@306: sort $WORK/$SYLLIST.tmp > $WORK/$SYLLIST
tomwalters@306: 
tomwalters@306: 
tomwalters@306: # Generate a list of filenames from the spoke pattern
tomwalters@306: cat <<"EOF" > $WORK/$TRAIN_SPEAKERS
tomwalters@306: 170.9p112.2s100.0t+000itd
tomwalters@306: 171.0p112.8s100.0t+000itd	
tomwalters@306: 171.3p111.7s100.0t+000itd	
tomwalters@306: 171.5p113.1s100.0t+000itd	
tomwalters@306: 171.9p111.5s100.0t+000itd	
tomwalters@306: 172.1p113.0s100.0t+000itd	
tomwalters@306: 172.4p111.9s100.0t+000itd	
tomwalters@306: 172.5p112.5s100.0t+000itd
tomwalters@306: EOF
tomwalters@306: 
tomwalters@306: cat <<"EOF" > $WORK/$TEST_SPEAKERS
tomwalters@306: 137.0p104.3s100.0t+000itd	
tomwalters@306: 141.3p135.4s100.0t+000itd	
tomwalters@306: 145.5p106.3s100.0t+000itd	
tomwalters@306: 148.8p128.8s100.0t+000itd	
tomwalters@306: 151.6p83.9s100.0t+000itd	
tomwalters@306: 153.0p108.1s100.0t+000itd	
tomwalters@306: 155.5p123.5s100.0t+000itd	
tomwalters@306: 156.7p90.6s100.0t+000itd	
tomwalters@306: 159.5p109.6s100.0t+000itd	
tomwalters@306: 161.1p119.4s100.0t+000itd	
tomwalters@306: 161.1p96.8s100.0t+000itd	
tomwalters@306: 163.4p157.6s100.0t+000itd	
tomwalters@306: 164.7p110.8s100.0t+000itd	
tomwalters@306: 164.9p102.1s100.0t+000itd	
tomwalters@306: 165.6p144.0s100.0t+000itd	
tomwalters@306: 165.7p116.2s100.0t+000itd	
tomwalters@306: 167.4p133.5s100.0t+000itd	
tomwalters@306: 167.8p106.5s100.0t+000itd	
tomwalters@306: 168.6p111.6s100.0t+000itd	
tomwalters@306: 168.9p125.4s100.0t+000itd	
tomwalters@306: 169.0p114.0s100.0t+000itd	
tomwalters@306: 170.0p109.7s100.0t+000itd	
tomwalters@306: 170.1p119.5s100.0t+000itd	
tomwalters@306: 171.0p115.5s100.0t+000itd	
tomwalters@306: 171.7p112.3s100.0t+000itd	
tomwalters@306: 172.4p109.3s100.0t+000itd	
tomwalters@306: 173.3p105.6s100.0t+000itd	
tomwalters@306: 173.5p115.0s100.0t+000itd	
tomwalters@306: 174.5p100.6s100.0t+000itd	
tomwalters@306: 174.5p110.6s100.0t+000itd	
tomwalters@306: 174.9p113.0s100.0t+000itd	
tomwalters@306: 175.7p118.5s100.0t+000itd	
tomwalters@306: 176.1p94.5s100.0t+000itd	
tomwalters@306: 178.0p108.5s100.0t+000itd	
tomwalters@306: 178.1p87.6s100.0t+000itd	
tomwalters@306: 178.8p123.6s100.0t+000itd	
tomwalters@306: 179.0p113.9s100.0t+000itd	
tomwalters@306: 180.4p80.1s100.0t+000itd	
tomwalters@306: 183.0p105.7s100.0t+000itd	
tomwalters@306: 183.0p130.4s100.0t+000itd	
tomwalters@306: 184.8p115.1s100.0t+000itd	
tomwalters@306: 188.1p139.2s100.0t+000itd	
tomwalters@306: 189.6p102.1s100.0t+000itd	
tomwalters@306: 192.7p116.7s100.0t+000itd	
tomwalters@306: 194.5p150.4s100.0t+000itd	
tomwalters@306: 198.1p97.9s100.0t+000itd	
tomwalters@306: 202.7p118.6s100.0t+000itd	
tomwalters@306: 208.6p93.2s100.0t+000itd	
tomwalters@306: 215.2p121.0s100.0t+000itd
tomwalters@306: EOF
tomwalters@306: 
tomwalters@306: # Construct the conversion scripts for AIMCopy (or HCopy) and 
tomwalters@306: # the master label files for the train and test sets
tomwalters@306: 
tomwalters@306: echo "Generating train and test scripts..."
tomwalters@306: if [ -a $WORK/$TRAIN_LIST ] 
tomwalters@306: then
tomwalters@306:   rm $WORK/$TRAIN_LIST
tomwalters@306: fi
tomwalters@306: if [ -a $WORK/$TEST_LIST ]
tomwalters@306: then
tomwalters@306:   rm $WORK/$TEST_LIST
tomwalters@306: fi  
tomwalters@306:  
tomwalters@306: exec 3> $WORK/$TRAIN_LIST
tomwalters@306: exec 5> $WORK/$TEST_LIST
tomwalters@306: for syllable in $(cat $WORK/$SYLLIST); do
tomwalters@306:   for speaker in $(cat $WORK/$TRAIN_SPEAKERS); do
tomwalters@306:     SOURCE_FILENAME=$SOUND_SOURCE/$syllable/${syllable}${speaker}.wav
tomwalters@306:     DEST_FILENAME=$WORK/$FEATURES_DIR/$syllable/${syllable}${speaker}
tomwalters@306:     echo "$SOURCE_FILENAME  ${DEST_FILENAME}.htk" >&3
tomwalters@306:   done
tomwalters@306:   for speaker in $(cat $WORK/$TEST_SPEAKERS); do
tomwalters@306:     SOURCE_FILENAME=$SOUND_SOURCE/$syllable/${syllable}${speaker}.wav
tomwalters@306:     DEST_FILENAME=$WORK/$FEATURES_DIR/$syllable/${syllable}${speaker}
tomwalters@306:     echo "$SOURCE_FILENAME  ${DEST_FILENAME}.htk" >&5
tomwalters@306:   done
tomwalters@306: done
tomwalters@306: exec 3>&-
tomwalters@306: exec 5>&-
tomwalters@306: 
tomwalters@306: # Make the necessary directories for the computed features
tomwalters@306: echo "Making directory structure..."
tomwalters@306: mkdir $WORK/$FEATURES_DIR
tomwalters@306: for syllable in $(cat $WORK/$SYLLIST); do
tomwalters@306:   mkdir $WORK/$FEATURES_DIR/$syllable
tomwalters@306: done
tomwalters@306: 
tomwalters@306: # Write the AIMCopy config file
tomwalters@306: echo "Creating AIMCopy config file..."
tomwalters@306: cat <<"EOF" > $WORK/$AIMCOPY_CONFIG
tomwalters@306: input.buffersize=480
tomwalters@306: sai.frame_period_ms=10.0
tomwalters@306: slice.normalize=true
tomwalters@306: ssi.weight_by_cutoff=false
tomwalters@306: ssi.weight_by_scaling=true
tomwalters@306: ssi.log_cycles_axis=true
tomwalters@306: EOF
tomwalters@306: echo "noise.level_db=$1" >> $WORK/$AIMCOPY_CONFIG
tomwalters@316: 
tomwalters@316: echo "Creating HCopy config file..."
tomwalters@316: cat <<"EOF" > $WORK/$HCOPY_CONFIG
tomwalters@316: # Coding parameters
tomwalters@316: SOURCEFORMAT= WAV
tomwalters@316: TARGETKIND = MFCC_0_D_A
tomwalters@316: TARGETRATE = 100000.0
tomwalters@316: SAVECOMPRESSED = T
tomwalters@316: SAVEWITHCRC = T
tomwalters@316: WINDOWSIZE = 250000.0
tomwalters@316: USEHAMMING = T
tomwalters@316: PREEMCOEF = 0.97
tomwalters@316: NUMCHANS = 200
tomwalters@316: CEPLIFTER = 22
tomwalters@316: NUMCEPS = 12
tomwalters@316: ENORMALISE = F
tomwalters@316: EOF
tomwalters@306:  
tomwalters@306: echo "Splitting data files..."
tomwalters@306: cat $WORK/${TRAIN_LIST} $WORK/${TEST_LIST} > $WORK/${COMBINED_LIST}
tomwalters@306: 
tomwalters@306: total_cores=$(($MACHINE_CORES))
tomwalters@306: echo -n $total_cores
tomwalters@306: echo " cores available"
tomwalters@306: total_files=`cat $WORK/${COMBINED_LIST} | wc -l | sed 's/ *//'`
tomwalters@306: echo -n $total_files
tomwalters@306: echo " files to process"
tomwalters@306: files_per_core=$(($total_files/$total_cores+1))
tomwalters@306: echo -n $files_per_core
tomwalters@306: echo " files per core"
tomwalters@306: split -l $files_per_core $WORK/${COMBINED_LIST} $WORK/split_list
tomwalters@306: splits=( $(ls $WORK/split_list*))
tomwalters@306: element=0
tomwalters@306: echo "Spawning tasks..."
tomwalters@306: for ((c=1;c<=$MACHINE_CORES;c+=1)); do
tomwalters@306:   s=${splits[$element]}
tomwalters@316:   if [ "$MFCC_FEATURES" ]
tomwalters@316:   then
tomwalters@316:     ${HTK_PREFIX}HCopy -T 1 -C $WORK/$HCOPY_CONFIG -S $s &
tomwalters@316:   else
tomwalters@316:     ${AIMCOPY_PREFIX}AIMCopy -C $WORK/$AIMCOPY_CONFIG -S $s &
tomwalters@316:   fi
tomwalters@306:   let element=element+1
tomwalters@306: done
tomwalters@306: 
tomwalters@306: echo "Waiting for tasks to complete..."
tomwalters@306: wait
tomwalters@306: echo "Done!"