bugs when doing "MODULE: 05 Build Trees"

[MaggieYou]

Hello，I did experiment following the"Robust group's Open Source Tutorial",things went smooth until I did the " 05 Build Trees", the command of win32 poped up "bldtree.exe has encountered a problem and needs to close. We are sorry for the inconvenience."

So ,who can tell me where the problem is?
Any answer will be appreciated.

[Ben Lau]

Ugh... now I'm really confused.

In between sending my report and getting your reply I updated by SVN checkout to an older revision to see if that would work. Then I updated again to try and reproduce the problem I was having and it doesn't happen anymore... :-( Maybe it was a problem with the version of SphinxTrain I checked out last month and then didn't have time to work on... except that I updated it earlier this morning and still had the problem. This is very strange.

So it is fixed already. I cleaned my SVN checkout and got all the code fresh. Compiled it and built the an4 models and everything seems to work.

Thanks for the quick reply.

Ben

[The Grand Janitor]

Maggie, it's nice to see you again.

Would you send us a log for step 5? -a

[Ben Lau]

I'm having this same problem on linux(Debian Sarge on Intel 2.6.8 kernel) I just recompiled SphinxTrain(r5911) and bldtree is segfaulting in scripts_pl/05.buildtrees/buildtree.pl so the trees aren't being created leading to the following error:
FATAL: "main.c", line 167: Unable to open /home/bmoore/Work/tutorial/an4/trees/an4.unpruned/AA-0.dtree for reading; No such file or directory
Something failed: ("/home/bmoore/Work/tutorial/an4/scripts_pl/06.prunetree/slave.state-tie-er.pl" -cfg "./etc/sphinx_train.cfg")

Here's the log from the bldtree run for the phone 'AA':
/home/bmoore/Work/tutorial/an4/bin/bldtree \ -meanfn /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/means \ -varfn /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/variances \ -treefn /home/bmoore/Work/tutorial/an4/trees/an4.unpruned/AA-0.dtree \ -moddeffn /home/bmoore/Work/tutorial/an4/model_architecture/an4.untied.mdef \ -mixwfn /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/mixture_weights \ -ts2cbfn .cont. \ -mwfloor 1e-30 \ -psetfn /home/bmoore/Work/tutorial/an4/model_architecture/an4.tree_questions \ -phone AA \ -state 0 \ -stwt 1.0 0.05 0.0 \ -ssplitmin 1 \ -ssplitmax 5 \ -ssplitthr 0 \ -csplitmin 1 \ -csplitmax 500 \ -csplitthr 0

[Switch] [Default] [Value]
-help no no
-example no no
-treefn /home/bmoore/Work/tutorial/an4/trees/an4.unpruned/AA-0.dtree
-moddeffn /home/bmoore/Work/tutorial/an4/model_architecture/an4.untied.mdef
-ts2cbfn .semi. .cont.
-meanfn /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/means
-varfn /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/variances
-fullvar no no
-varfloor 0.00001 1.000000e-05
-cntthresh 0.00001 1.000000e-05
-mixwfn /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/mixture_weights
-psetfn /home/bmoore/Work/tutorial/an4/model_architecture/an4.tree_questions
-phone AA
-allphones no no
-state 0
-mwfloor 1e-4 1.000000e-30
-stwt 1.0 0.05 0.0
-ssplitthr 8e-4 0.000000e+00
-ssplitmin 1 1
-ssplitmax 5 5
-csplitthr 8e-4 0.000000e+00
-csplitmin 1 1
-csplitmax 100 500
INFO: main.c(228): Reading: /home/bmoore/Work/tutorial/an4/model_architecture/an4.untied.mdef
INFO: model_def_io.c(587): Model definition info:
INFO: model_def_io.c(588): 1618 total models defined (34 base, 1584 tri)
INFO: model_def_io.c(589): 6472 total states
INFO: model_def_io.c(590): 4854 total tied states
INFO: model_def_io.c(591): 102 total tied CI states
INFO: model_def_io.c(592): 34 total tied transition matrices
INFO: model_def_io.c(593): 4 max state/model
INFO: model_def_io.c(594): 4 min state/model
INFO: main.c(247): Building trees for [AA AH R b] through [AA UW R b]
INFO: main.c(275): Covering states |[102 161]| == 60
INFO: main.c(282): Reading: /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/mixture_weights
INFO: s3mixw_io.c(172): Read /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/mixture_weights [60x1x1 array]
INFO: main.c(320): 20 of 20 models have observation count greater than 0.000010
INFO: main.c(94): nrm stwt: 0.952 0.048 0.000
INFO: s3gau_io.c(162): Read /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/means [4854x1x1 array]
INFO: s3gau_io.c(162): Read /home/bmoore/Work/tutorial/an4/model_parameters/an4.cd_cont_untied/variances [4854x1x1 array]
Thu Jul 20 15:25:24 2006

If I run the command line at the top manually it shows the Segmentation Fault.

Any help would be much appreciated.

Ben

[The Grand Janitor]

Hi Ben,
This is rather strange. Maggies's bug is a windows issue. rev 5911 has already fixed that so I am very curious to fix the one you mentioned. so please help us on this,

1, could you run the command-line with valgrind? What you need to do is to
a) download and install valgrind from valgrind.org.
b) type
>valgrind <the command you typed> > log 2>&1

Send us the log file and I will try to take a look of it asap.

-a