[hts-users:04607] ERROR at embedded reestimation (fullcontext) step

Hi All,

i'm training HMM voice using MARY TTS, which run HTS procedure "make data". . The HMMVoiceMakeData procedure is similar to the original HTS scripts with additional sections for calculating voicing strengths (str), for mixed excitation, and composing training data files from mgc, lf0 and str files. This component executes in the hts/data/ directory:

cd training/voice/path/
cd hts
perl scripts/Training.pl scripts/Config.pm > logfile &

i have recently encountered a problem while running 'embedded reestimation (fullcontext)' :

'''

==============================================================================================
Start embedded reestimation (fullcontext) at الخميس حزيران 21 11:17:36 EEST 2018
==============================================================================================

Embedded Re-estimation
/home/hiyassat/Trining_EqraFemale1Voice/marytts-5.1/lib/external/bin/HERest -A -B -C /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/configs/trn.cnf -D -T 1 -S /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/data/scp/train.scp -I /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/data/labels/full.mlf -m 1 -u tmvwdmv -w 5000 -t 1500 100 5000 -H /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/models/qst001/ver1/cmp/fullcontext.mmf -N /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/models/qst001/ver1/dur/fullcontext.mmf -M /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/models/qst001/ver1/cmp -R /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/models/qst001/ver1/dur -C /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/configs/nvf.cnf -s /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/stats/qst001/ver1/cmp.stats -w 0.0 /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/data/lists/full.list /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/data/lists/full.list

'''

killed

'''

Error in /home/hiyassat/Trining_EqraFemale1Voice/marytts-5.1/lib/external/bin/HERest -A -B -C /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/configs/trn.cnf -D -T 1 -S /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/data/scp/train.scp -I /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/data/labels/full.mlf -m 1 -u tmvwdmv -w 5000 -t 1500 100 5000 -H /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/models/qst001/ver1/cmp/fullcontext.mmf -N /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/models/qst001/ver1/dur/fullcontext.mmf -M /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/models/qst001/ver1/cmp -R /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/models/qst001/ver1/dur -C /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/configs/nvf.cnf -s /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/stats/qst001/ver1/cmp.stats -w 0.0 /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/data/lists/full.list /media/hiyassat/Data/Training_EqraFemale1Voice/training_all/Eqra_female1_abqriat_voice_2018/hts/data/lists/full.list

'''

Can anyone tell what the problem might be? i didn't had any problem with small database(up to 2.5 hour audio), it was done successfully. but this time i am using huge data (28 hour audio) , when 'embedded reestimation (fullcontext) step started, my computer freezing and not responding then the error appear in terminal, so is it a memory problem or what?. Kindly anyone please guide me.

i have attached HTS log files and Training script.

Regards

,doaa obiedat

log files hts