tomwalters@54: #!/bin/bash
tomwalters@54: # Script to run a single HMM train/test cycle with the given parameters.
tomwalters@54: # This script expects the following variables to be set
tomwalters@54: #
tomwalters@54: # total_hmm_states - total number of HMM states (including the 2 non-emitting states)
tomwalters@54: # mixture_components - number of components in the output distribution for each emitting state
tomwalters@54: # input_vector_size - number or elements in the input vector (normally 39 for MFCCs, 12 for AIM)
tomwalters@54: # feature_code - HTK feature type code for the features being used (normally MFCC_0_D_A for MFCCs and USER_E_D_A for AIM features)
tomwalters@54: # FEATURE_SUFFIXES - List of suffixes appended to the feature filenames. For the MFCCs this is just "mfc" but for the AIM feature, there can be multiple features generated from each run of AIMCopy
tomwalters@54: # WORK - working directory
tomwalters@54: # SYLLIST_COMPLETE
tomwalters@54: 
tomwalters@54: # Filenames generated here
tomwalters@54: HMMPROTO=hmm_prototype
tomwalters@54: HHED_SCRIPT=hhed_change_components_script
tomwalters@54: RECOUT=recognition_output
tomwalters@54: RESULTS_FILE=results
tomwalters@54: MISCLASSIFIED=misclassified_syllables
tomwalters@54: 
tomwalters@54: # Filenames used here
tomwalters@54: TRAIN_SCRIPT=training_script
tomwalters@54: TEST_SCRIPT=testing_script
tomwalters@54: SYLLIST_COMPLETE=syllable_list_with_silence
tomwalters@54: 
tomwalters@54: hmm_type=${total_hmm_states}states_${mixture_components}mixture_components
tomwalters@54: echo "HMM type: $hmm_type"
tomwalters@54: mkdir -p $WORK/$hmm_type
tomwalters@54: 
tomwalters@54: echo "Creating HMM structure..."
tomwalters@54: ./gen_hmmproto.py --input_size ${input_vector_size} --total_hmm_states ${total_hmm_states} --feature_type ${feature_code} > $WORK/$hmm_type/$HMMPROTO
tomwalters@54: 
tomwalters@54: echo "Adding output mixture components..."
tomwalters@54: ./gen_hhed_script.py --num_means ${mixture_components} --total_hmm_states ${total_hmm_states} > $WORK/$hmm_type/$HHED_SCRIPT
tomwalters@54: 
tomwalters@54: 
tomwalters@54: echo "Training HMM..."
tomwalters@54: echo "Setting up prototype HMM..."
tomwalters@54: mkdir -p $WORK/$hmm_type/hmm0
tomwalters@54: HCompV -C $WORK/$HMMCONFIG -f 0.01 -m -S $WORK/$TRAIN_SCRIPT -M $WORK/$hmm_type/hmm0 $WORK/$hmm_type/$HMMPROTO
tomwalters@54: 
tomwalters@54: echo "Generating HMM definitions..."
tomwalters@54: # Now take the prototype file from hmm0, and create the other HMM definitions
tomwalters@54: # from it
tomwalters@54: grep -A 9999 "<BEGINHMM>" $WORK/$hmm_type/hmm0/$HMMPROTO > $WORK/$hmm_type/hmm0/hmms
tomwalters@54: for syllable in $(cat $WORK/$SYLLIST_COMPLETE); do
tomwalters@54:   echo "~h $syllable" >> $WORK/$hmm_type/$feature/hmm0/hmmdefs
tomwalters@54:   cat $WORK/$hmm_type/$feature/hmm0/hmms >> $WORK/$hmm_type/$feature/hmm0/hmmdefs
tomwalters@54: done
tomwalters@54:  
tomwalters@54: echo -n "~o<STREAMINFO> 1 ${input_vector_size}<VECSIZE> ${input_vector_size}<NULLD><${feature_code}><DIAGC>" > $WORK/$hmm_type/$feature/hmm0/macros
tomwalters@54: 
tomwalters@54: cat $WORK/$hmm_type/$feature/hmm0/vFloors >> $WORK/$hmm_type/$feature/hmm0/macros
tomwalters@54: 
tomwalters@54: ${HTK_PREFIX}HHEd  -H $WORK/$hmm_type/$feature/hmm0/macros -H $WORK/$hmm_type/$feature/hmm0/hmmdefs $WORK/$hmm_type/$HHED_SCRIPT $WORK/$SYLLIST_COMPLETE
tomwalters@54: 
tomwalters@54: for iter in $TRAINING_ITERATIONS_LIST; do
tomwalters@54:   echo "Training iteration ${iter}..."
tomwalters@54:   let "nextiter=$iter+1"
tomwalters@54:   mkdir $WORK/$hmm_type/hmm$nextiter
tomwalters@54:   ${HTK_PREFIX}HERest -C $WORK/$HMMCONFIG -I $WORK/$TRAIN_MLF \
tomwalters@54:     -t 250.0 150.0 1000.0 -S $WORK/$TRAIN_SCRIPT \
tomwalters@54:     -H $WORK/$hmm_type/hmm$iter/macros -H $WORK/$hmm_type/hmm$iter/hmmdefs \
tomwalters@54:     -M $WORK/$hmm_type/hmm$nextiter $WORK/$SYLLIST_COMPLETE
tomwalters@54: done
tomwalters@54: 
tomwalters@54: for iter in $TESTING_ITERATIONS_LIST; do
tomwalters@54:   echo "Testing iteration ${iter}..."
tomwalters@54:   ${HTK_PREFIX}HVite -H $WORK/$hmm_type/hmm$iter/macros -H $WORK/$hmm_type/hmm$iter/hmmdefs \
tomwalters@54:     -C $WORK/$HMMCONFIG -S $WORK/$TEST_SCRIPT -i $WORK/$hmm_type/$RECOUT \
tomwalters@54:     -w $WORK/$WDNET -p 0.0 -s 5.0 $WORK/$DICT $WORK/$SYLLIST_COMPLETE
tomwalters@54:   echo "Results from testing on iteration ${iter}..."
tomwalters@54:   ${HTK_PREFIX}HResults -e "???" ${SILENCE} -I $WORK/$TEST_MLF $WORK/$SYLLIST_COMPLETE $WORK/$hmm_type/$RECOUT
tomwalters@54:   ${HTK_PREFIX}HResults -p -t -e "???" ${SILENCE} \
tomwalters@54:     -I $WORK/$TEST_MLF $WORK/$SYLLIST_COMPLETE $WORK/$hmm_type/$RECOUT > $WORK/$hmm_type/${RESULTS_FILE}_iteration_$iter
tomwalters@54:   grep Aligned $WORK/$hmm_type/${RESULTS_FILE}_iteration_$iter| sed -E "s/.*\/..\/([a-z]{2})([0-9]{2,3}\.[0-9])p([0-9]{2,3}\.[0-9])s.*/\2 \3/" | sort | uniq -c > $WORK/$hmm_type/${MISCLASSIFIED}_iteration_$iter
tomwalters@54: done