README

Step 01: Extract WORLD features

python 001_extract_world_feats.py 
./bin/AbS_world ascii_to_float_world feats_world files.head

Step 02: Build base voice and save the linguistic features

# Set voice directory
mkdir cmu_us_rsk
cd cmu_us_rsk
$FESTVOXDIR/src/clustergen cmu us rsk

# Copy the wavefiles and text file 
./bin/get_wavs /home/srallaba/data/tts/arctic/rsk/wav/arctic*
cp /home/srallaba/data/tts/arctic/rsk/txt.done.data.arctic > etc/txt.done.data

# Set to save the features and coefficients
cp festvox/clustergen.scm festvox/clustergen.scm.xxx
cat festvox/clustergen.scm.xxx | sed 's/(defvar cg:save_dumped_coeff_files nil)/(defvar cg:save_dumped_coeff_files t)/' > festvox/clustergen.scm

# Text preprocessing
./bin/do_build build_prompts etc/txt.done.data
./bin/do_build label etc/txt.done.data
./bin/do_clustergen parallel build_utts etc/txt.done.data
./bin/do_clustergen generate_statenames etc/txt.done.data
./bin/do_clustergen generate_filters etc/txt.done.data

# Extract features
./bin/do_clustergen parallel f0_v_sptk etc/txt.done.data
./bin/do_clustergen parallel mcep_sptk etc/txt.done.data
./bin/do_clustergen parallel combine_coeffs_v etc/txt.done.data
 
 # Build a base voice
./bin/traintest etc/txt.done.data
./bin/do_clustergen parallel cluster etc/txt.done.data.train
./bin/do_clustergen dur etc/txt.done.data.train
 
 cd ..
 
Step 03: Dump linguistic features
sh 03_001_make_featstoDNN.sh
sh 03_002_make_bin_feats.sh

Step 04: Adjust the coefficients based on EHMM labels
./bin/AbS_world  combine_coeffs_utt feats_world files

Step 05 : Map !!
sh map.sh