FATAL_ERROR: "corpus.c" duri...

[Pezhman Lali]

Dear
The Runall.pl script makes the following error. I can not find the reason by
the googling, may be you can help me

sphinxtrain 1.0.7
pocketsphinx 0.7
base 0.7

./scripts_pl/RunAll.pl

MODULE: 00 verify training files
O.S. is case sensitive ("A" != "a").
Phones will be treated as case sensitive.
Phase 1: DICT - Checking to see if the dict and filler dict agrees with the
phonelist file.
Found 6 words using 14 phones
Phase 2: DICT - Checking to make sure there are not duplicate entries in the
dictionary
Phase 3: CTL - Check general format; utterance length (must be positive);
files exist
Phase 4: CTL - Checking number of lines in the transcript should match lines
in control file
Phase 5: CTL - Determine amount of training data, see if n_tied_states seems
reasonable.
Estimated Total Hours Training: 0.00261666666666667
This is a small amount of data, no comment at this time
Phase 6: TRANSCRIPT - Checking that all the words in the transcript are in the
dictionary
Words in dictionary: 3
Words in filler dictionary: 3
Phase 7: TRANSCRIPT - Checking that all the phones in the transcript are in
the phonelist, and all phones in the phonelist appear at least once
Feature type is s2_4x which is 4 streams
LDA/MLLT only has sense for single stream features, for example 1s_c_d_dd
Skipping LDA training
Feature type is s2_4x which is 4 streams
LDA/MLLT only has sense for single stream features, for example 1s_c_d_dd
Skipping MLLT training
MODULE: 05 Vector Quantization
This step had 2 ERROR messages and 3191 WARNING messages. Please check the log
file for details.
MODULE: 10 Training Context Independent models for forced alignment and VTLN
Skipped: $ST::CFG_FORCEDALIGN set to 'no' in sphinx_train.cfg
Skipped: $ST::CFG_VTLN set to 'no' in sphinx_train.cfg
MODULE: 11 Force-aligning transcripts
Skipped: $ST::CFG_FORCEDALIGN set to 'no' in sphinx_train.cfg
MODULE: 12 Force-aligning data for VTLN
Skipped: $ST::CFG_VTLN set to 'no' in sphinx_train.cfg
MODULE: 20 Training Context Independent models
Phase 1: Cleaning up directories:
accumulator...logs...qmanager...models...
Phase 2: Flat initialize
Phase 3: Forward-Backward
Training failed in iteration 1
Something failed: (/root/sphinx/man/scripts_pl/20.ci_hmm/slave_convg.pl)

from the log :

==> man.1.1-2.bw.log <==
INFO: main.c(397): Will reestimate means.
INFO: main.c(399): Will reestimate variances.
INFO: main.c(407): Will reestimate transition matrices
INFO: main.c(420): Reading main lexicon: /root/sphinx/man/etc/man.dic
INFO: lexicon.c(233): 3 entries added from /root/sphinx/man/etc/man.dic
INFO: main.c(432): Reading filler lexicon: /root/sphinx/man/etc/man.filler
INFO: lexicon.c(233): 3 entries added from /root/sphinx/man/etc/man.filler
INFO: corpus.c(436): skipping 4 utts.
FATAL_ERROR: "corpus.c", line 1345: File length mismatch at line 4 in
/root/sphinx/man/etc/man_train.transcription
Tue Dec 13 06:40:18 2011

[Pezhman Lali]

this is /root/sphinx/man/etc/man_train.transcription

cat /root/sphinx/man/etc/man_train.transcription

FAARSI (file_1)
ENGHILIYSI (file_2)
BAAZDIYD (file_3)

[Nickolay V. Shmyrev]

Your transcription file has less lines than your fileids file.

Your database is too small for training

For more information on training read the tutorial

http://cmusphinx.sourceforge.net/wiki/tutorialam

[Pezhman Lali]

Thanks for your reply

This is the fileid

cat man_train.fileids

TOM1/file_1
TOM1/file_2
TOM1/file_3
TOM2/file_1
TOM2/file_2
TOM2/file_3
Mary1/file_1
Mary1/file_2
Mary1/file_3

The file id has more lines, Because we have only 3 words for recognition, but
we have some speakers(TOM1,TOM2,Tom3, Mary1,Mary2,.... am I right ?

[Nickolay V. Shmyrev]

Transcription lines for each speaker should be present. The lines should
repeat. The number of lines in fileids file must be equal to the number of
lines in transcription file.

I definitely suggest you to go through the tutorial first with the test an4
database. Then do by analogy.