I am starting training new model with MMIE is on, I have copied the 2 files (Sphinx3_align and Sphinx3_decode ) .
training was stopped at MODULE: 61 Lattice Pruning
Phase 1: Cleaning up directories ...
logs...qmanager...pruned_denlat...
Phase 2: Posterior probability lattice pruning ...
with the following error
ERROR: "ngram_model_arpa.c", line 155: No \data\ mark in LM file
ERROR: "ngram_model_dmp.c", line 117: Wrong magic header size number 54726965: /media/Data/an4/etc/tmp/an4.lm.DMP is not a dump file
process sent: muratel/001002
load lattice ...
Traceback (most recent call last):
File "/usr/local/lib/sphinxtrain/python/cmusphinx/lattice_prune.py", line 65, in <module>
dag.edges_unigram_score(lm,lw)
File "/usr/local/lib/sphinxtrain/python/cmusphinx/lattice.py", line 1192, in edges_unigram_score
e.lscr = lm.prob([baseword(e.src.sym)]) * lw
File "sphinxbase.pyx", line 155, in sphinxbase.NGramModel.prob
TypeError: expected string or Unicode object, list found
Mon Apr 1 15:58:00 2019</module>
i tried all suggestions by other users of sphinx on the web
even i uninstall all sphinxbase,sphinxtrain and pocketsphinx and install fresh instalation of https://cmusphinx.github.io/wiki/download/
i installed
sphinxbase-5prealpha
pocketsphinx-5prealpha
sphinxtrain-5prealpha
but the problem is still exist
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
now the MODULE: 62 Lattice Format Conversion Phase 2: Converting lattice format . collapsed with this error
/usr/local/lib/sphinxtrain/python/cmusphinx/lattice_conv.py /media/Data/an4/etc/tmp/an4.lm.DMP /media/Data/an4/etc/an4.filler /media/Data/an4/numlat/an4.alignedfiles 34697 242879 /media/Data/an4/pruned_denlat /media/Data/an4/numlat /media/Data/an4/lattice
Traceback (most recent call last):
File "/usr/local/lib/sphinxtrain/python/cmusphinx/lattice_conv.py", line 284, in <module>
filler = load_lexicon(fillerfile)
File "/usr/local/lib/sphinxtrain/python/cmusphinx/lattice_conv.py", line 29, in load_lexicon
lex[temp[0]] = temp[1:len(temp)]
IndexError: list index out of range
Tue Apr 2 14:50:25 2019</module>
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I am starting training new model with MMIE is on, I have copied the 2 files (Sphinx3_align and Sphinx3_decode ) .
training was stopped at MODULE: 61 Lattice Pruning
Phase 1: Cleaning up directories ...
logs...qmanager...pruned_denlat...
Phase 2: Posterior probability lattice pruning ...
with the following error
ERROR: "ngram_model_arpa.c", line 155: No \data\ mark in LM file
ERROR: "ngram_model_dmp.c", line 117: Wrong magic header size number 54726965: /media/Data/an4/etc/tmp/an4.lm.DMP is not a dump file
process sent: muratel/001002
load lattice ...
Traceback (most recent call last):
File "/usr/local/lib/sphinxtrain/python/cmusphinx/lattice_prune.py", line 65, in <module>
dag.edges_unigram_score(lm,lw)
File "/usr/local/lib/sphinxtrain/python/cmusphinx/lattice.py", line 1192, in edges_unigram_score
e.lscr = lm.prob([baseword(e.src.sym)]) * lw
File "sphinxbase.pyx", line 155, in sphinxbase.NGramModel.prob
TypeError: expected string or Unicode object, list found
Mon Apr 1 15:58:00 2019</module>
i tried all suggestions by other users of sphinx on the web
even i uninstall all sphinxbase,sphinxtrain and pocketsphinx and install fresh instalation of https://cmusphinx.github.io/wiki/download/
i installed
sphinxbase-5prealpha
pocketsphinx-5prealpha
sphinxtrain-5prealpha
but the problem is still exist
the problem solved
i download sphinxbase 8, sphinxtrain 8 and build the dmp file using the new genertated sphinx_lm_convert tool
and problem solved
now the MODULE: 62 Lattice Format Conversion Phase 2: Converting lattice format . collapsed with this error
/usr/local/lib/sphinxtrain/python/cmusphinx/lattice_conv.py /media/Data/an4/etc/tmp/an4.lm.DMP /media/Data/an4/etc/an4.filler /media/Data/an4/numlat/an4.alignedfiles 34697 242879 /media/Data/an4/pruned_denlat /media/Data/an4/numlat /media/Data/an4/lattice
Traceback (most recent call last):
File "/usr/local/lib/sphinxtrain/python/cmusphinx/lattice_conv.py", line 284, in <module>
filler = load_lexicon(fillerfile)
File "/usr/local/lib/sphinxtrain/python/cmusphinx/lattice_conv.py", line 29, in load_lexicon
lex[temp[0]] = temp[1:len(temp)]
IndexError: list index out of range
Tue Apr 2 14:50:25 2019</module>
the problem solved
there are empty line at the end of filler dict