Error: Fstheader::read: Bad Fst Header: Standard Input
Carte ▾ ▴
Mistake running mkgraph
2014-07-30
2014-07-31
-
-
Probable there was a problem when creating L_disambig.fst. It should be
dumped without symbols (the options --keep_isymbols=false
--keep_osymbols=false in the scripts).Dan
On Tue, Jul 29, 2014 at 9:13 PM, enzyme156 enzyme156@users.sf.cyberspace wrote:
Hello,
After successfully doing monophone training, I am moving on to making the
decoding graph. However running mkgraph.sh gave this mistake:fsttablecompose information/lang/L_disambig.fst data/lang/G.fst
WARNING: CompatSymbols: first symbol table present only second missing
FATAL: ComposeFst: output symbol tabular array of 1st argument does not match
input symbol tabular array of 2nd argument
fstminimizeencoded
fstdeterminizestar --use-log=truthful
Fault: FstHeader::Read: Bad FST header: -
ERROR (fstdeterminizestar:ReadFstKaldi():fstext/fstext-utils-inl.h:1184)
Reading FST: error reading FST header from standard input
Error (fstdeterminizestar:ReadFstKaldi():fstext/fstext-utils-inl.h:1184)
Reading FST: error reading FST header from standard input[stack trace: ]
0 fstdeterminizestar 0x000000010c6ae074 _ZN5kaldi18KaldiGetStackTraceEv +
54
i fstdeterminizestar 0x000000010c6aecde _ZN5kaldi17KaldiErrorMessageD1Ev +
122
ii fstdeterminizestar 0x000000010c661cff _ZN3fst12ReadFstKaldiESs + 315
three fstdeterminizestar 0x000000010c65f11b chief + 883
iv libdyld.dylib 0x00007fff912c15fd start + 1
5 ??? 0x0000000000000002 0x0 + 2ERROR: FstHeader::Read: Bad FST header: -
ERROR (fstminimizeencoded:ReadFstKaldi():fstext/fstext-utils-inl.h:1184)
Reading FST: fault reading FST header from standard input
ERROR (fstminimizeencoded:ReadFstKaldi():fstext/fstext-utils-inl.h:1184)
Reading FST: error reading FST header from standard input[stack trace: ]
0 fstminimizeencoded 0x000000010ab563ae _ZN5kaldi18KaldiGetStackTraceEv +
54
ane fstminimizeencoded 0x000000010ab57018 _ZN5kaldi17KaldiErrorMessageD1Ev +
122
2 fstminimizeencoded 0x000000010ab2259b _ZN3fst12ReadFstKaldiESs + 315
three fstminimizeencoded 0x000000010ab2033c main + 384
4 libdyld.dylib 0x00007fff912c15fd start + 1Whatever idea what might be the trouble? Give thanks you!
Error running mkgraph
https://sourceforge.net/p/kaldi/discussion/1355348/thread/fc92bd75/?limit=25#b854
Sent from sourceforge.cyberspace because you indicated involvement in
https://sourceforge.cyberspace/p/kaldi/discussion/1355348/To unsubscribe from further messages, delight visit
https://sourceforge.net/auth/subscriptions/If you would like to refer to this comment somewhere else in this projection, copy and paste the following link:
-
Are y'all talking almost when the lang directory was generated? Should I regenerate the L_disambig.fst file somehow?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
-
Hello Daniel,
After using fstsymbols to remove symbols from L_disambig.fst, I got this mistake instead:
fsttablecompose data/lang/L_disambig.fst data/lang/G.fst
fstminimizeencoded
Mistake: FstHeader::Read: Bad FST header: data/lang/L_disambig.fst
Fault (fsttablecompose:ReadFstKaldi():fstext/fstext-utils-inl.h:1184) Reading FST: error reading FST header from data/lang/L_disambig.fst
fstdeterminizestar --use-log=true
Error (fsttablecompose:ReadFstKaldi():fstext/fstext-utils-inl.h:1184) Reading FST: mistake reading FST header from data/lang/L_disambig.fstUsing validate_lang.pl on the lang directory gave this mistake:
Checking word_boundary.int and disambig.int
--> generating a 32 words sequence
--> resulting telephone sequence from 50.fst corresponds to the give-and-take sequence
--> L.fst is OK
--> generating a 100 words sequence
ERROR: FstHeader::Read: Bad FST header: information/lang/L_disambig.fst
Fault: FstHeader::Read: Bad FST header: standard input
Error: FstHeader::Read: Bad FST header: standard input
ERROR: FstHeader::Read: Bad FST header: standard input
ERROR: FstHeader::Read: Bad FST header: standard input
Fault: FstHeader::Read: Bad FST header: standard input
--> Fault: number of reconstructed words 0 does not lucifer real number of words 100; indicates problem in L_disambig.fst or word_boundary.int. phoneseq =If you would similar to refer to this comment somewhere else in this projection, copy and paste the following link:
-
It looks like you accidentally replaced L_disambig.fst with the empty file.
If yous do
fstrmsymbols <foo>foo
it won't work, you need to use a temporary file.
DanOn Wed, Jul 30, 2014 at 12:07 AM, enzyme156 enzyme156@users.sf.cyberspace wrote:
Hi Daniel,
After using fstsymbols to remove symbols from L_disambig.fst, I got this
mistake instead:fsttablecompose data/lang/L_disambig.fst information/lang/G.fst
fstminimizeencoded
Fault: FstHeader::Read: Bad FST header: information/lang/L_disambig.fst
ERROR (fsttablecompose:ReadFstKaldi():fstext/fstext-utils-inl.h:1184)
Reading FST: error reading FST header from information/lang/L_disambig.fst
fstdeterminizestar --use-log=true
Mistake (fsttablecompose:ReadFstKaldi():fstext/fstext-utils-inl.h:1184)
Reading FST: error reading FST header from data/lang/L_disambig.fstUsing validate_lang.pl on the lang directory gave this fault:
Checking word_boundary.int and disambig.int
--> generating a 32 words sequence
--> resulting phone sequence from L.fst corresponds to the give-and-take sequence
--> L.fst is OK
--> generating a 100 words sequence
Error: FstHeader::Read: Bad FST header: data/lang/L_disambig.fst
Mistake: FstHeader::Read: Bad FST header: standard input
Error: FstHeader::Read: Bad FST header: standard input
ERROR: FstHeader::Read: Bad FST header: standard input
ERROR: FstHeader::Read: Bad FST header: standard input
Mistake: FstHeader::Read: Bad FST header: standard input
--> Error: number of reconstructed words 0 does non match real number of
words 100; indicates problem in L_disambig.fst or word_boundary.int.
phoneseq =
Error running mkgraph
https://sourceforge.net/p/kaldi/word/1355348/thread/fc92bd75/?limit=25#d4b7
Sent from sourceforge.net because you indicated interest in
https://sourceforge.net/p/kaldi/discussion/1355348/To unsubscribe from farther letters, delight visit
https://sourceforge.net/auth/subscriptions/If you would similar to refer to this comment somewhere else in this project, copy and paste the post-obit link:
-
Hello Daniel,
I tried fstrmsymbols with G.fst and that got mkgraph working.
Withal I ran into a problem when training triphones.
This is the console output:
steps/train_deltas.sh: line 106: 3888 Abort trap: six cluster-phones $dir/treeacc $lang/phones/sets.int $dir/questions.int 2> $dir/log/questions.log
And this is the questions.log file:
cluster-phones exp/tri1/treeacc data/lang/phones/sets.int exp/tri1/questions.int
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580) Phone eleven is present in stats only is not in phone list [make sure you intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580) Phone 13 is nowadays in stats just is not in phone list [make sure you lot intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580) Telephone 15 is present in stats but is not in phone list [make sure you intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580) Telephone xx is present in stats but is not in phone listing [make sure you intended this].
Warning (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580) Phone 23 is present in stats but is not in phone listing [make sure you intended this].
etc...
etc..
etc...
Alarm (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580) Phone 147 is present in stats but is not in telephone list [make sure you lot intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:610) All or all but one of your classes of phones had no information. Note that we just consider data where pdf-grade is in the set ( ane ). If you take an unusual HMM topology this may not be what y'all desire; use the --pdf-class-listing pick to change this if needed. Encounter also whatever warnings to a higher place.
KALDI_ASSERT: at cluster-phones:ObtainSetsOfPhones:build-tree.cc:472, failed: num_leaves < static_cast<int32>(clust_assignments.size())I checked my sets.int and the contents of that file was:
i
2
3
iv
5I tried generating a phone.listing file to be used instead of sets.int, but got this error in build_tree.log:
WARNING (build-tree:main():build-tree.cc:160) Mismatch between telephone sets provided in roots file, and those in topology: [ one two 3 4 5 ]
vs. [ 1 2 3 iv five vi 7 8 ix x eleven 12 13 fourteen xv 16 17 eighteen 19 twenty 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 twoscore 41 42 43 44 45 46 47 48 49 fifty 51 52 53 54 55 56 57 58 59 threescore 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 ]
Mistake (build-tree:primary():build-tree.cc:171) Phone 11 appears in stats just is non listed in roots file.
ERROR (build-tree:main():build-tree.cc:171) Phone eleven appears in stats simply is not listed in roots file.Practise you accept whatsoever ideas what the trouble is? Thank y'all very much!
If you would similar to refer to this comment somewhere else in this project, re-create and paste the following link:
-
There was a problem with your data preparation, your sets.int should
contain all phones. Try validate_data_dir.sh. If your data directory was
prepared using utils/prepare_lang.sh information technology should be OK.
DanOn Thu, Jul 31, 2014 at 12:21 AM, enzyme156 enzyme156@users.sf.net wrote:
Hi Daniel,
I tried fstrmsymbols with Yard.fst and that got mkgraph working.
Yet I ran into a problem when training triphones.
This is the console output:
steps/train_deltas.sh: line 106: 3888 Abort trap: 6 cluster-phones
$dir/treeacc $lang/phones/sets.int $dir/questions.int ii>
$dir/log/questions.logAnd this is the questions.log file:
cluster-phones exp/tri1/treeacc data/lang/phones/sets.int exp/tri1/
questions.int
Alarm (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone eleven is present in stats but is not in phone listing [brand certain you
intended this].
Alarm (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone 13 is nowadays in stats simply is not in phone list [make sure you
intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone 15 is present in stats but is not in phone listing [make sure yous
intended this].
Warning (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone 20 is present in stats but is non in phone list [brand sure y'all
intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone 23 is nowadays in stats simply is not in telephone list [brand certain yous
intended this].
etc...
etc..
etc...
Alarm (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone 147 is present in stats but is not in phone listing [make certain you
intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:610)
All or all only one of your classes of phones had no data. Note that we only
consider data where pdf-class is in the set up ( i ). If you have an unusual
HMM topology this may non be what y'all want; use the --pdf-class-list option
to modify this if needed. Run across besides any warnings above.
KALDI_ASSERT: at cluster-phones:ObtainSetsOfPhones:build-tree.cc:472,
failed: num_leaves < static_cast<int32>(clust_assignments.size())I checked my sets.int and the contents of that file was:
1
2
3
4
vI tried generating a phone.list file to exist used instead of sets.int, but
got this error in build_tree.log:Alarm (build-tree:principal():build-tree.cc:160) Mismatch betwixt phone sets
provided in roots file, and those in topology: [ 1 two 3 four 5 ]
vs. [ 1 two 3 iv 5 half-dozen 7 8 nine x 11 12 13 14 xv 16 17 eighteen 19 20 21 22 23 24 25
26 27 28 29 thirty 31 32 33 34 35 36 37 38 39 forty 41 42 43 44 45 46 47 48 49 50
51 52 53 54 55 56 57 58 59 sixty 61 62 63 64 65 66 67 68 69 seventy 71 72 73 74 75
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138
139 140 141 142 143 144 145 146 147 148 149 150 ]
ERROR (build-tree:main():build-tree.cc:171) Telephone 11 appears in stats but
is non listed in roots file.
ERROR (build-tree:main():build-tree.cc:171) Phone eleven appears in stats but
is not listed in roots file.Do you accept any ideas what the problem is? Thanks very much!
Fault running mkgraph
https://sourceforge.internet/p/kaldi/discussion/1355348/thread/fc92bd75/?limit=25#d4b7/d8d8/ef69
Sent from sourceforge.net because you indicated interest in
https://sourceforge.cyberspace/p/kaldi/discussion/1355348/To unsubscribe from further messages, please visit
https://sourceforge.net/auth/subscriptions/If y'all would like to refer to this comment somewhere else in this project, copy and paste the following link:
-
I mean
utils/validate_lang.sh data/lang
(or something like that)
to check your lang/ directory. but the sets.int file should incorporate all
the phones that you have.
DanOn Thu, Jul 31, 2014 at 12:23 AM, Daniel Povey dpovey@gmail.com wrote:
There was a problem with your information preparation, your sets.int should
contain all phones. Try validate_data_dir.sh. If your data directory was
prepared using utils/prepare_lang.sh it should be OK.
DanOn Thu, Jul 31, 2014 at 12:21 AM, enzyme156 enzyme156@users.sf.internet
wrote:How-do-you-do Daniel,
I tried fstrmsymbols with Yard.fst and that got mkgraph working.
However I ran into a trouble when training triphones.
This is the console output:
steps/train_deltas.sh: line 106: 3888 Abort trap: 6 cluster-phones
$dir/treeacc $lang/phones/sets.int $dir/questions.int 2>
$dir/log/questions.logAnd this is the questions.log file:
cluster-phones exp/tri1/treeacc data/lang/phones/sets.int exp/tri1/
questions.int
Alarm (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone 11 is present in stats but is not in telephone listing [brand sure you lot
intended this].
Alarm (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone xiii is present in stats but is not in phone list [make sure you lot
intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Telephone 15 is present in stats but is not in phone list [make sure you
intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone 20 is nowadays in stats but is not in phone listing [brand sure y'all
intended this].
Alarm (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone 23 is present in stats but is non in telephone list [make certain you lot
intended this].
etc...
etc..
etc...
Warning (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:580)
Phone 147 is present in stats but is not in phone listing [make sure you lot
intended this].
WARNING (cluster-phones:AutomaticallyObtainQuestions():build-tree.cc:610)
All or all but one of your classes of phones had no data. Note that we only
consider information where pdf-course is in the gear up ( ane ). If yous have an unusual
HMM topology this may non be what you desire; utilize the --pdf-class-list selection
to alter this if needed. Run across also any warnings higher up.
KALDI_ASSERT: at cluster-phones:ObtainSetsOfPhones:build-tree.cc:472,
failed: num_leaves < static_cast<int32>(clust_assignments.size())I checked my sets.int and the contents of that file was:
one
2
3
4
5I tried generating a phone.list file to be used instead of sets.int, but
got this error in build_tree.log:Alert (build-tree:principal():build-tree.cc:160) Mismatch between telephone sets
provided in roots file, and those in topology: [ 1 2 3 4 5 ]
vs. [ i 2 3 four 5 6 7 8 9 ten xi 12 13 fourteen fifteen 16 17 xviii nineteen twenty 21 22 23 24 25
26 27 28 29 30 31 32 33 34 35 36 37 38 39 twoscore 41 42 43 44 45 46 47 48 49 50
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138
139 140 141 142 143 144 145 146 147 148 149 150 ]
ERROR (build-tree:main():build-tree.cc:171) Phone xi appears in stats merely
is not listed in roots file.
ERROR (build-tree:primary():build-tree.cc:171) Phone xi appears in stats but
is non listed in roots file.Do you take any ideas what the problem is? Thank you very much!
Fault running mkgraph
https://sourceforge.net/p/kaldi/discussion/1355348/thread/fc92bd75/?limit=25#d4b7/d8d8/ef69
Sent from sourceforge.net because y'all indicated interest in
https://sourceforge.net/p/kaldi/word/1355348/To unsubscribe from farther letters, please visit
https://sourceforge.net/auth/subscriptions/If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
-
Hi Daniel,
validate_lang.sh returned SUCCESS for me.
Even when replacing sets.int with a phone.list file (which contains all of my phones) I still bumped into the error above when doing build_trees. It seems like the root.int file did not contain all of the phones either.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
-
It looks like validate_lang.pl does non cheque that roots.int contains all
the phones.
But it should contain all of them.
sets.int should at least contain all of the non-silence phones. I doubtable
you did your own data preparation and did not exercise information technology quite right. If you
utilize the prepare_lang.sh script it should give you lot a correct upshot; if not,
send me the input directory yous used to prepare_lang.sh at dpovey@gmail.com,
and the resulting output lang/ directory, then I tin can meet what went wrong.
DanOn Thu, Jul 31, 2014 at 12:31 AM, enzyme156 enzyme156@users.sf.cyberspace wrote:
Hullo Daniel,
validate_lang.sh returned SUCCESS for me.
Fifty-fifty when replacing sets.int with a phone.list file (which contains all
of my phones) I still bumped into the error higher up when doing build_trees.
It seems like the root.int file did not contain all of the phones either.
Mistake running mkgraph
https://sourceforge.cyberspace/p/kaldi/discussion/1355348/thread/fc92bd75/?limit=25#d4b7/d8d8/ef69/0210/57a8/e869
Sent from sourceforge.cyberspace because you indicated involvement in
https://sourceforge.net/p/kaldi/give-and-take/1355348/To unsubscribe from further messages, please visit
https://sourceforge.net/auth/subscriptions/If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
-
It turned out that this trouble was due to some files in the original dict/
directory given to prepare_lang.sh non ending in newlines. I fixed
validate_dict_dir.sh to detect this and dice.
If someone has time, information technology would be not bad if they could update validate_lang.pl
to ensure that roots.int is valid in the sense that it contains all phones
(except disambiguation symbols, of course) and does not contain any
repeated phones on unlike lines. People tin let me know at
dpovey@gmail.com if they accept fourth dimension for this.
DanOn Thu, Jul 31, 2014 at 12:35 AM, Daniel Povey danielpovey@users.sf.net
wrote:It looks like validate_lang.pl does non cheque that roots.int contains all
the phones.
Merely information technology should contain all of them.
sets.int should at least contain all of the non-silence phones. I suspect
you lot did your own data preparation and did not practice it quite right. If you lot
utilise the prepare_lang.sh script it should give you lot a correct upshot; if not,
send me the input directory you used to prepare_lang.sh at
dpovey@gmail.com,
and the resulting output lang/ directory, so I tin can see what went wrong.
DanOn Thu, Jul 31, 2014 at 12:31 AM, enzyme156 enzyme156@users.sf.net wrote:
Hello Daniel,
validate_lang.sh returned SUCCESS for me.
Fifty-fifty when replacing sets.int with a telephone.list file (which contains all
of my phones) I notwithstanding bumped into the fault above when doing build_trees.
It seems like the root.int file did not comprise all of the phones either.
Error running mkgraph
https://sourceforge.net/p/kaldi/word/1355348/thread/fc92bd75/?limit=25#d4b7/d8d8/ef69/0210/57a8/e869
Sent from sourceforge.net because you indicated interest in
https://sourceforge.net/p/kaldi/discussion/1355348/To unsubscribe from further messages, delight visit
https://sourceforge.cyberspace/auth/subscriptions/
Fault running mkgraph
http://sourceforge.internet/p/kaldi/discussion/1355348/thread/fc92bd75/?limit=25#d4b7/d8d8/ef69/0210/57a8/e869/5c23
Sent from sourceforge.net considering yous indicated involvement in
https://sourceforge.net/p/kaldi/give-and-take/1355348/To unsubscribe from further messages, please visit
https://sourceforge.net/auth/subscriptions/If you would similar to refer to this comment somewhere else in this project, copy and paste the post-obit link:
Source: https://sourceforge.net/p/kaldi/discussion/1355348/thread/fc92bd75/
Hello,
After successfully doing monophone preparation, I am moving on to making the decoding graph. Even so running mkgraph.sh gave this error:
[stack trace: ]
0 fstdeterminizestar 0x000000010c6ae074 _ZN5kaldi18KaldiGetStackTraceEv + 54
1 fstdeterminizestar 0x000000010c6aecde _ZN5kaldi17KaldiErrorMessageD1Ev + 122
ii fstdeterminizestar 0x000000010c661cff _ZN3fst12ReadFstKaldiESs + 315
3 fstdeterminizestar 0x000000010c65f11b principal + 883
iv libdyld.dylib 0x00007fff912c15fd start + ane
five ??? 0x0000000000000002 0x0 + 2
Fault: FstHeader::Read: Bad FST header: -
Error (fstminimizeencoded:ReadFstKaldi():fstext/fstext-utils-inl.h:1184) Reading FST: error reading FST header from standard input
Mistake (fstminimizeencoded:ReadFstKaldi():fstext/fstext-utils-inl.h:1184) Reading FST: error reading FST header from standard input
[stack trace: ]
0 fstminimizeencoded 0x000000010ab563ae _ZN5kaldi18KaldiGetStackTraceEv + 54
1 fstminimizeencoded 0x000000010ab57018 _ZN5kaldi17KaldiErrorMessageD1Ev + 122
2 fstminimizeencoded 0x000000010ab2259b _ZN3fst12ReadFstKaldiESs + 315
3 fstminimizeencoded 0x000000010ab2033c chief + 384
4 libdyld.dylib 0x00007fff912c15fd kickoff + 1
Whatsoever thought what might exist the trouble? Thank you!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link: