cmusphinx/sphinxtrain

Acoustic model trainer for CMU Sphinx

sphinxtrain's Issues

scripts/decode/word_align.py is not python3-ready

It mixes tabs and spaces, uses print foo

sphinx3_align not available

sphinx3_align is used for force aligning in scripts/11.force_align/force_align.pl, scripts/11.force_align/slave_align.pl, etc. But sphinx3_align itself is not available in sphinxtrain after sphinx3, and there doesn't exist any good approach to get it (Building old version sphinx3 with old version sphinxbase meets many other problems).

Any good ideas to solve this problem? And I think stating it clearly in wiki/tutorialam will be better.

error when executing the autogen in ubuntu

./configure: line 14407: syntax error near unexpected token `use_lapack=true'
./configure: line 14407: `use_lapack=true'

I have tried to figure out the issue by spending many times but without any success.

Restarting PTM training in stage 20 fails

Something isn't right with the way the initial model gets saved, because if you kill and restart stage 20 you get this:

ERROR: FATAL: "main.c", line 338: src mean n_cb (== 40) is inconsistent with ts2cb mapping 1. Most probably phoneset has duplicated phones

Split jobs according to time rather than number of utterances

Decoding in particular wastes a lot of time waiting for parts of the data that happen to contain very long utterances.

The simpler alternative is to shuffle the data, though this has its own problems.

Add slurm support

While various PBS-like schedulers still live on, it seems that most people use Slurm these days, and there's even a relatively straightforward tutorial on setting it up on Google Compute Engine ... which looks significantly more helpful than the "documentation" for CycleCloud on Azure.

Elasticloud still exists too, it seems to perhaps support slurm better than others.

So we should support slurm, it should be pretty easy given the small subset of scheduler functionality we need.

bw.exe breaks on windows

I'm on windows trying to adapt the generic data modules. I've cloned the repository and built the project using visual studio. I use bw.exe with the following command and part of the way through execution I get a popup error message that says "bw.exe has stopped working \n A problem caused the program to stop working correctly. Windows will close the program and notify you if a solution is available".

What's the deal? I am on 64 bit version of windows. Does that not work with this?

training is not properly restartable

If running sphinxtrain train and it fails (as it does, because the Perl scripts have bugs, but also as it might if using spot instances on $CLOUD) then there's no easy way to restart training.

Back in the old days we would just sit in our offices at CMU all night running scripts_pl/NN.step/s***ve_confg.pl manually, but I would prefer to be in the forest picking mushrooms these days.

This isn't rocket science, at the very least it could just restart from the step and iteration, though of course, it would be much better to rerun just the parts that failed. We're not even using GPUs so there's no issue with repeatability when doing that.

problem in new acoustic model

ohk. I used en-us.lm.bin file and then i executed this command " sphinx_fe -argfile en-us/feat.params-samprate 16000 -c arctic20.fileids -di . -do . -ei wav -eo mfc -mswav yes"

.mfc files are generated but there is no change in cmudict-en-us.dict and en-us.lm.bin files . The sentences which i recorded where i can see them. How i will access my acoustic model.
please help.

macOS compilation error: lexicon.c:202:12: error: implicit declaration of function 'str2words' is invalid in C99

I am unable to compile the code on macOS 12.1 Monterey. After running ./configure, the make commands fails:

Making all in libcommon
/bin/sh ../../../libtool  --tag=CC   --mode=compile gcc -DHAVE_CONFIG_H -I. -I../../../include   -I/Users/lorenzo/Downloads/sphinx_install/sphinxtrain/../sphinxbase/include -I/Users/lorenzo/Downloads/sphinx_install/sphinxtrain/../sphinxbase/include  -I../../../include -g -O2 -fPIC -DPIC -MT lexicon.lo -MD -MP -MF .deps/lexicon.Tpo -c -o lexicon.lo lexicon.c
libtool: compile:  gcc -DHAVE_CONFIG_H -I. -I../../../include -I/Users/lorenzo/Downloads/sphinx_install/sphinxtrain/../sphinxbase/include -I/Users/lorenzo/Downloads/sphinx_install/sphinxtrain/../sphinxbase/include -I../../../include -g -O2 -fPIC -DPIC -MT lexicon.lo -MD -MP -MF .deps/lexicon.Tpo -c lexicon.c  -fno-common -DPIC -o .libs/lexicon.o
lexicon.c:202:12: error: implicit declaration of function 'str2words' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
        n_phone = str2words(line->buf, NULL, 0) - 1;
                  ^
1 error generated.
make[3]: *** [lexicon.lo] Error 1
make[2]: *** [all-recursive] Error 1
make[1]: *** [all-recursive] Error 1
make: *** [all-recursive] Error 1

I have both Perl v5.30.3 (which is installed by default on macOS) and Python 3.9.7 installed. I also installed the latest version (5-prealpha) of sphinxbase and pocketsphinx through code compilation (sudo make install) and I have the compiled source code of both in folders named sphinxbase and pocketsphinx respectively in the same parent folder as my sphinxtrain folder I am trying to compile. (The absolute path to those folders does not contain any spaces, as this seem to be an issue as well.)

Finally, adding #include <sphinxbase/strfuncs.h> to src/libs/libcommon/lexicon.c and src/libs/libcommon/mk_phone_seq.c seems to solve the problem, however another error occurs:

Making all in libio
/bin/sh ../../../libtool  --tag=CC   --mode=compile gcc -DHAVE_CONFIG_H -I. -I../../../include   -I/Users/lorenzo/Downloads/sphinx_install/sphinxtrain/../sphinxbase/include -I/Users/lorenzo/Downloads/sphinx_install/sphinxtrain/../sphinxbase/include  -I../../../include -g -O2 -fPIC -DPIC -MT corpus.lo -MD -MP -MF .deps/corpus.Tpo -c -o corpus.lo corpus.c
libtool: compile:  gcc -DHAVE_CONFIG_H -I. -I../../../include -I/Users/lorenzo/Downloads/sphinx_install/sphinxtrain/../sphinxbase/include -I/Users/lorenzo/Downloads/sphinx_install/sphinxtrain/../sphinxbase/include -I../../../include -g -O2 -fPIC -DPIC -MT corpus.lo -MD -MP -MF .deps/corpus.Tpo -c corpus.c  -fno-common -DPIC -o .libs/corpus.o
corpus.c:1314:12: error: implicit declaration of function 'areadfloat' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
            ret = areadfloat(mk_filename(DATA_TYPE_MFCC, cur_ctl_path),
                  ^
corpus.c:1318:12: error: implicit declaration of function 'areadfloat_part' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
            ret = areadfloat_part(mk_filename(DATA_TYPE_MFCC, cur_ctl_path),
                  ^
corpus.c:1388:9: error: implicit declaration of function 'areadshort' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
    if (areadshort(mk_filename(DATA_TYPE_SEG, rel_path), (int16**)seg, n_seg) < 0)
        ^
3 errors generated.
make[3]: *** [corpus.lo] Error 1
make[2]: *** [all-recursive] Error 1
make[1]: *** [all-recursive] Error 1
make: *** [all-recursive] Error 1

I suppose adding another include would solve this problem as well but I guess this is not the way to do it. Am I missing something ?

scripts have sketchy names

While it's unlikely that anyone would seriously be triggered or offended by a script called slave_convg.pl it's true that we wouldn't call it, say, concentration_camp.pl or residential_school.pl.

Beyond being in bad taste, it just doesn't really mean anything useful. We won't be removing Nietzsche's "Beyond Good and Evil" from the LibriSpeech development set or anything like that.

We could call it "Håkan" but perhaps a better name would be something like "train_from_iteration.pl" since, well, that's what it does.

lexicon.c:202:12: error: implicit declaration of function 'str2words' is invalid

I had ran "./configure" arelady. And than ran "make" but not successfully.

$ make
Making all in etc
make[1]: Nothing to be done for all'. Making all in include /Applications/Xcode.app/Contents/Developer/usr/bin/make all-am Making all in python make[1]: Nothing to be done for all'.
Making all in scripts
make[1]: Nothing to be done for all'. Making all in src Making all in libs Making all in libclust make[3]: Nothing to be done for all'.
Making all in libcommon
/bin/sh ../../../libtool --tag=CC --mode=compile gcc -DHAVE_CONFIG_H -I. -I../../../include -I/Users/Aaron/iOS/Sphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha-已编译/include -I/Users/Aaron/iOS/Sphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha-已编译/include -I../../../include -g -O2 -fPIC -DPIC -MT lexicon.lo -MD -MP -MF .deps/lexicon.Tpo -c -o lexicon.lo lexicon.c
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../include -I/Users/Aaron/iOS/Sphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha-已编译/include -I/Users/Aaron/iOS/Sphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha-已编译/include -I../../../include -g -O2 -fPIC -DPIC -MT lexicon.lo -MD -MP -MF .deps/lexicon.Tpo -c lexicon.c -fno-common -DPIC -o .libs/lexicon.o
lexicon.c:202:12: error: implicit declaration of function 'str2words' is invalid
in C99 [-Werror,-Wimplicit-function-declaration]
n_phone = str2words(line->buf, NULL, 0) - 1;
^
1 error generated.
make[3]: *** [lexicon.lo] Error 1
make[2]: *** [all-recursive] Error 1
make[1]: *** [all-recursive] Error 1
make: *** [all-recursive] Error 1

Bug in makefile

On Ubuntu 12.04, i have cloned this repo and run automake.sh.
When it successfully ended i run make and it ends up with an error, that i had no .libs/libsphinxbase.so.

libtool: link: gcc -I../../../include -g -O2 -fPIC -DPIC -o .libs/agg_seg agg_all_seg.o agg_phn_seg.o agg_st_seg.o cnt_phn_seg.o cnt_st_seg.o main.o mk_seg.o parse_cmd_ln.o  -LMYHOMEDIR/sphinxtrain/../sphinxbase/src/libsphinxbase/.libs -LMYHOMEDIR/sphinxtrain/../sphinxbase/src/libsphinxad -LMYHOMEDIR/sphinxtrain/../sphinxbase/src/libsphinxbase ../../../src/libs/libio/.libs/libio.a ../../../src/libs/libmodinv/.libs/libmodinv.a ../../../src/libs/libcommon/.libs/libcommon.a ../../../src/libs/libclust/.libs/libclust.a ../../../src/libs/libmllr/.libs/libmllr.a -lpthread -lm MYHOMEDIR/sphinxtrain/../sphinxbase/src/libsphinxbase/.libs/libsphinxbase.so
gcc: error: MYHOMEDIR/sphinxtrain/../sphinxbase/src/libsphinxbase/.libs/libsphinxbase.so: No such file or directory

Directory contents are
libsphinxbase (2).la -> ../libsphinxbase.la
libsphinxbase (2).so -> libsphinxbase.so.1.1.1
libsphinxbase (3).so
libsphinxbase (4).so
libsphinxbase (5).so
libsphinxbase (6).so
libsphinxbase.a
libsphinxbase.lai
libsphinxbase.so.1.1.1
libsphinxbase.so (2).1 -> libsphinxbase.so.1.1.1

When i try to copy libsphinxbase.so.1.1.1 with name libsphinxbase.so. after running make i get another libsphinxbase (i).so file and an error gcc: error: MYHOMEDIR/sphinxtrain/../sphinxbase/src/libsphinxbase/.libs/libsphinxbase.so: No such file or directory

sphinxtrain fails to make on raspberry pi 2

log:

shell
~/sphinxtrain $ make
Making all in etc
make[1]: Entering directory '/home/pi/sphinxtrain/etc'
make[1]: Nothing to be done for 'all'.
make[1]: Leaving directory '/home/pi/sphinxtrain/etc'
Making all in include
make[1]: Entering directory '/home/pi/sphinxtrain/include'
make[1]: Nothing to be done for 'all'.
make[1]: Leaving directory '/home/pi/sphinxtrain/include'
Making all in python
make[1]: Entering directory '/home/pi/sphinxtrain/python'
make[1]: Nothing to be done for 'all'.
make[1]: Leaving directory '/home/pi/sphinxtrain/python'
Making all in scripts
make[1]: Entering directory '/home/pi/sphinxtrain/scripts'
make[1]: Nothing to be done for 'all'.
make[1]: Leaving directory '/home/pi/sphinxtrain/scripts'
Making all in src
make[1]: Entering directory '/home/pi/sphinxtrain/src'
Making all in libs
make[2]: Entering directory '/home/pi/sphinxtrain/src/libs'
Making all in libclust
make[3]: Entering directory '/home/pi/sphinxtrain/src/libs/libclust'
make[3]: Nothing to be done for 'all'.
make[3]: Leaving directory '/home/pi/sphinxtrain/src/libs/libclust'
Making all in libcommon
make[3]: Entering directory '/home/pi/sphinxtrain/src/libs/libcommon'
make[3]: Nothing to be done for 'all'.
make[3]: Leaving directory '/home/pi/sphinxtrain/src/libs/libcommon'
Making all in libio
make[3]: Entering directory '/home/pi/sphinxtrain/src/libs/libio'
make[3]: Nothing to be done for 'all'.
make[3]: Leaving directory '/home/pi/sphinxtrain/src/libs/libio'
Making all in libmllr
make[3]: Entering directory '/home/pi/sphinxtrain/src/libs/libmllr'
make[3]: Nothing to be done for 'all'.
make[3]: Leaving directory '/home/pi/sphinxtrain/src/libs/libmllr'
Making all in libmodinv
make[3]: Entering directory '/home/pi/sphinxtrain/src/libs/libmodinv'
make[3]: Nothing to be done for 'all'.
make[3]: Leaving directory '/home/pi/sphinxtrain/src/libs/libmodinv'
make[3]: Entering directory '/home/pi/sphinxtrain/src/libs'
make[3]: Nothing to be done for 'all-am'.
make[3]: Leaving directory '/home/pi/sphinxtrain/src/libs'
make[2]: Leaving directory '/home/pi/sphinxtrain/src/libs'
Making all in programs
make[2]: Entering directory '/home/pi/sphinxtrain/src/programs'
Making all in agg_seg
make[3]: Entering directory '/home/pi/sphinxtrain/src/programs/agg_seg'
/bin/bash ../../../libtool --tag=CC --mode=link gcc -I../../../include -g -O2 -L/home/pi/sphinxtrain/../sphinxbase-0.8/src/libsphinxad -L/home/pi/sphinxtrain/../sphinxbase-0.8/src/libsphinxbase -L/home/pi/sphinxtrain/../sphinxbase-0.8/src/libsphinxbase/.libs -o agg_seg agg_all_seg.o agg_phn_seg.o agg_st_seg.o cnt_phn_seg.o cnt_st_seg.o main.o mk_seg.o parse_cmd_ln.o ../../../src/libs/libio/libio.la ../../../src/libs/libmodinv/libmodinv.la ../../../src/libs/libcommon/libcommon.la ../../../src/libs/libclust/libclust.la ../../../src/libs/libmllr/libmllr.la -lm -lsphinxbase
libtool: link: gcc -I../../../include -g -O2 -o .libs/agg_seg agg_all_seg.o agg_phn_seg.o agg_st_seg.o cnt_phn_seg.o cnt_st_seg.o main.o mk_seg.o parse_cmd_ln.o -L/home/pi/sphinxtrain/../sphinxbase-0.8/src/libsphinxbase/.libs -L/home/pi/sphinxtrain/../sphinxbase-0.8/src/libsphinxad -L/home/pi/sphinxtrain/../sphinxbase-0.8/src/libsphinxbase ../../../src/libs/libio/.libs/libio.a ../../../src/libs/libmodinv/.libs/libmodinv.a ../../../src/libs/libcommon/.libs/libcommon.a ../../../src/libs/libclust/.libs/libclust.a ../../../src/libs/libmllr/.libs/libmllr.a -lasound -lpthread -lm /home/pi/sphinxtrain/../sphinxbase-0.8/src/libsphinxbase/.libs/libsphinxbase.so
../../../src/libs/libio/.libs/libio.a(s3map_io.o): In function s3map_write': /home/pi/sphinxtrain/src/libs/libio/s3map_io.c:152: undefined reference to bio_fwrite_1d'
collect2: ld returned 1 exit status
Makefile:336: recipe for target 'agg_seg' failed
make[3]: *** [agg_seg] Error 1
make[3]: Leaving directory '/home/pi/sphinxtrain/src/programs/agg_seg'
Makefile:321: recipe for target 'all-recursive' failed
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory '/home/pi/sphinxtrain/src/programs'
Makefile:287: recipe for target 'all-recursive' failed
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/home/pi/sphinxtrain/src'
Makefile:379: recipe for target 'all-recursive' failed
make: *** [all-recursive] Error 1

cannont find .dmp find

hi,
i follow the following tutorial
https://cmusphinx.github.io/wiki/tutorialadapt/
In "Adapting the acoustic model " section , it is said to copy en-us.lm.dmp file from source location to my current working directory. But en-us.lm.dmp file is not there in the source location . I have en-us.lm.bin , en-us-phone.lm.bin, cmudict-en-us.dict files . Is there any alternate option that can be used in place of en-us.lm.dmp file .
please help

-remove_noise not included in feat.params

Since a change to PocketSphinx this is considered (as it should be) part of the feature extraction parameters, so SphinxTrain should put it there.

Step "Creating a transformation with MLLR", log out so many "INFO: mllr.c(186): Estimation of 0 th regression in MLLR failed".

so many "INFO: mllr.c(186): Estimation of 0 th regression in MLLR failed" out long with adapting, and use the adapted result, the test result accuracy is so poor. What's the problem? Lagu is Mandarin. tks.

How to install it on Windows 10 ?

Hello everyone,

I am struggling with installing sphinxtrain for several days already. As I am a newbie in the IT game I might completely misunderstand how to install the toolkit.

What I have done so far :

VS Code :
Version: 1.77.1 (user setup)
Commit: b7886d7461186a5eac768481578c1d7ca80e2d21
Date: 2023-04-04T23:21:11.906Z
Electron: 19.1.11
Chromium: 102.0.5005.196
Node.js: 16.14.2
V8: 10.2.154.26-electron.0
OS: Windows_NT x64 10.0.19044
Sandboxed: No

Extensions :
C/C++ v1.14.5
CMake Tools v1.13.45

CMake installed version 3.26.3, (correct PATH registered in WINDOWS variable environment)

Python 3.11.3
Perl 5.36.0

I am currently trying to execute "cmake -S . -B build" in the VSCode terminal. Unfortunately the answer in red says that "cmake is not recognised..."

+ cmake -S . -B build
+ ~~~~~
    + CategoryInfo          : ObjectNotFound: (cmake:String) [], 

I also tried in the CMD terminal and that time :

CMake Error at CMakeLists.txt:3 (project):
  Running

   'nmake' '-?'

  failed with:

   Le fichier spécifié est introuvable


CMake Error: CMAKE_C_COMPILER not set, after EnableLanguage
CMake Error: CMAKE_CXX_COMPILER not set, after EnableLanguage
-- Configuring incomplete, errors occurred!

Am I supposed to run this command in a Terminal, or am I supposed to already have the ".sln" file ??

Thank you in advance for your time and consideration,

Best regards,

Francis

Want to set mfc file location in "bw" on commandline

Hi,

I want to set "mfc" file location in "bw" on commandline.so is there any flag for that?
Actually I wrote one script for all adaption command, and my acoustic model location is diff than my script location. I thought "-cepext" will work for me but no it is for mfc extension only.

Thanks in advance.

semi/ptm training is excessively, probably unnecessarily slow

In theory, semi-continuous or PTM models are supposed to be fast! But training them is incredibly slow, especially the initial flat start. This is most likely due to some redundant or inefficient computation in the training code.

Training a 128-Gaussian PTM model on 100 hours of data on 16 CPUs takes approximately 4 hours, whereas training a 4000 senone, 16 Gaussian continuous model with LDA and MLLT takes only 1h25 (without LDA and MLLT it's less than an hour).

And of course the accuracy of said PTM model is quite atrocious.

One might argue that they are thoroughly obsolete, but actually they are maybe the only reason left to use CMU Sphinx, since they produce very small models.

compilation crashes because of slave_feat.pl

everytime I try to train my model I get this message:

compilation failed in require at /usr/local/lib/sphinxtrain/scripts/000.comp_feat/slave_feat.pl line 51.
BEGIN failed--compilation aborted at /usr/local/lib/sphinxtrain/scripts/000.comp_feat/slave_feat.pl line 51

anyone have some ideia of what this is??

cannot find the corresponding hmm during "sphinxtrain run"

I got an error that
MODULE: DECODE Decoding using models previously trained
Decoding 3 segments starting at 0 (part 1 of 1)
0% ERROR: FATAL: "batch.c", line 822: PocketSphinx decoder init failed

ERROR: This step had 1 ERROR messages and 0 WARNING messages. Please check the log file for details.
ERROR: Failed to start pocketsphinx_batch
Aligning results to find error rate
Can't open ./test_train_sphinx/result/去你媽的擔擔麵-1-1.match
word_align.pl failed with error code 65280 at /usr/local/lib/sphinxtrain/scripts/decode/slave.pl line 173.

So, I checked th the log in logdir/decode/去你媽的擔擔麵.log
I saw the line

-hmm ./test_train_sphinx/model_parameters/去你媽的擔擔麵.cd_cont_200

The substring _200 is generated by

Models to use.

$DEC_CFG_MODEL_NAME = "$CFG_EXPTNAME.cd_${CFG_DIRLABEL}_${CFG_N_TIED_STATES}";

in sphinxt_train.cfg

IS IT A BUG???????

Sphinxtrain make failed on virtualbox and ubuntu : Error in making libclust

Hello,
i have try another sphnxtrain installation on virtualbox and ubuntu , my previous was on Ubuntu for windows. And this time i have the same problem on making sphinxtrain with error on making libclust.

I use the last libtool 2.4.6 and 18.04.1-Ubuntu
sphinxtrain was the last version like sphinxbase and pocketsphinx.

=========== make sphinxtrain ======

thierry@vmLFR038391:~/pocketsphinx/sphinxtrain-5prealpha$ make
Making all in etc
make[1] : on entre dans le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/etc »
make[1]: rien à faire pour « all ».
make[1] : on quitte le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/etc »
Making all in include
make[1] : on entre dans le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/include »
(CDPATH="${ZSH_VERSION+.}:" && cd .. && /bin/bash /home/thierry/pocketsphinx/sphinxtrain-5prealpha/missing autoheader)
rm -f stamp-h1
touch config.h.in
cd .. && /bin/bash ./config.status include/config.h
config.status: creating include/config.h
config.status: include/config.h is unchanged
make all-am
make[2] : on entre dans le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/include »
make[2] : on quitte le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/include »
make[1] : on quitte le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/include »
Making all in python
make[1] : on entre dans le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/python »
make[1]: rien à faire pour « all ».
make[1] : on quitte le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/python »
Making all in scripts
make[1] : on entre dans le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/scripts »
make[1]: rien à faire pour « all ».
make[1] : on quitte le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/scripts »
Making all in src
make[1] : on entre dans le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/src »
Making all in libs
make[2] : on entre dans le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/src/libs »
Making all in libclust
make[3] : on entre dans le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/src/libs/libclust »
/bin/bash ../../../libtool --tag=CC --mode=compile gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT div.lo -MD -MP -MF .deps/div.Tpo -c -o div.lo div.c
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT div.lo -MD -MP -MF .deps/div.Tpo -c div.c -fPIC -DPIC -o .libs/div.o
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT div.lo -MD -MP -MF .deps/div.Tpo -c div.c -o div.o >/dev/null 2>&1
mv -f .deps/div.Tpo .deps/div.Plo
/bin/bash ../../../libtool --tag=CC --mode=compile gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT kdtree.lo -MD -MP -MF .deps/kdtree.Tpo -c -o kdtree.lo kdtree.c
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT kdtree.lo -MD -MP -MF .deps/kdtree.Tpo -c kdtree.c -fPIC -DPIC -o .libs/kdtree.o
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT kdtree.lo -MD -MP -MF .deps/kdtree.Tpo -c kdtree.c -o kdtree.o >/dev/null 2>&1
mv -f .deps/kdtree.Tpo .deps/kdtree.Plo
/bin/bash ../../../libtool --tag=CC --mode=compile gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT kmeans.lo -MD -MP -MF .deps/kmeans.Tpo -c -o kmeans.lo kmeans.c
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT kmeans.lo -MD -MP -MF .deps/kmeans.Tpo -c kmeans.c -fPIC -DPIC -o .libs/kmeans.o
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT kmeans.lo -MD -MP -MF .deps/kmeans.Tpo -c kmeans.c -o kmeans.o >/dev/null 2>&1
mv -f .deps/kmeans.Tpo .deps/kmeans.Plo
/bin/bash ../../../libtool --tag=CC --mode=compile gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT metric.lo -MD -MP -MF .deps/metric.Tpo -c -o metric.lo metric.c
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT metric.lo -MD -MP -MF .deps/metric.Tpo -c metric.c -fPIC -DPIC -o .libs/metric.o
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/include -I../../../include -g -O2 -fPIC -DPIC -MT metric.lo -MD -MP -MF .deps/metric.Tpo -c metric.c -o metric.o >/dev/null 2>&1
mv -f .deps/metric.Tpo .deps/metric.Plo
/bin/bash ../../../libtool --tag=CC --mode=link gcc -I../../../include -g -O2 -fPIC -DPIC -L/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/src/libsphinxad -L/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/src/libsphinxbase -L/home/thierry/pocketsphinx/sphinxtrain-5prealpha/../sphinxbase-5prealpha/src/libsphinxbase/.libs -o libclust.la div.lo kdtree.lo kmeans.lo metric.lo -lm -lsphinxbase
../../../libtool: eval: line 7977: unexpected EOF while looking for matching `''
../../../libtool: eval: line 7978: syntax error: unexpected end of file
Makefile:357: recipe for target 'libclust.la' failed
make[3]: [libclust.la] Error 2
make[3] : on quitte le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/src/libs/libclust »
Makefile:361: recipe for target 'all-recursive' failed
make[2]: [all-recursive] Error 1
make[2] : on quitte le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/src/libs »
Makefile:361: recipe for target 'all-recursive' failed
make[1]: [all-recursive] Error 1
make[1] : on quitte le répertoire « /home/thierry/pocketsphinx/sphinxtrain-5prealpha/src »
Makefile:457: recipe for target 'all-recursive' failed
make: [all-recursive] Error 1

==========================================

it's seems a libtool problem perhaps same as in this topic :
https://bugs.launchpad.net/ubuntu/+source/libtool/+bug/1002565

PLease help me in this situation, i am completely lost !!!

map_adapt segfault

Doing the adaptation tutorial: http://cmusphinx.sourceforge.net/wiki/tutorialadapt?s[]=adaptation

Finally get near the payoff, something like:

./map_adapt \
    -meanfn model/means \
    -varfn model/variances \
    -mixwfn model/mixture_weights \
    -tmatfn model/transition_matrices \
    -accumdir . \ 
    -mapmeanfn model-adapt/means \
    -mapvarfn model-adapt/variances \
    -mapmixwfn model-adapt/mixture_weights \
    -maptmatfn model-adapt/transition_matrices

segfaults

INFO: cmd_ln.c(697): Parsing command line:
./map_adapt \
    -meanfn model/means \
    -varfn model/variances \
    -mixwfn model/mixture_weights \
    -tmatfn model/transition_matrices \
    -accumdir . \
    -mapmeanfn model-adapt/means \
    -mapvarfn model-adapt/variances \
    -mapmixwfn model-adapt/mixture_weights \
    -maptmatfn model-adapt/transition_matrices 

Current configuration:
[NAME]      [DEFLT] [VALUE]
-accumdir       .,
-bayesmean  yes yes
-example    no  no
-fixedtau   no  no
-help       no  no
-mapmeanfn      model-adapt/means
-mapmixwfn      model-adapt/mixture_weights
-maptmatfn      model-adapt/transition_matrices
-mapvarfn       model-adapt/variances
-meanfn         model/means
-mixwfn         model/mixture_weights
-mwfloor    0.00001 1.000000e-05
-tau        10.0    1.000000e+01
-tmatfn         model/transition_matrices
-tpfloor    0.0001  1.000000e-04
-varfloor   0.00001 1.000000e-05
-varfn          model/variances

INFO: s3gau_io.c(169): Read model/means [42x3x128 array]
INFO: s3gau_io.c(169): Read model/variances [42x3x128 array]
INFO: s3mixw_io.c(117): Read model/mixture_weights [5126x3x128 array]
INFO: s3tmat_io.c(118): Read model/transition_matrices [42x3x4 array]
INFO: main.c(426): Reading and accumulating observation counts from .
INFO: s3gau_io.c(386): Read ./gauden_counts with means with vars [42x3x128 vector arrays]
INFO: s3mixw_io.c(117): Read ./mixw_counts [5126x3x128 array]
INFO: s3tmat_io.c(118): Read ./tmat_counts [42x3x4 array]
INFO: main.c(77): Estimating tau hyperparameter from variances and observations
do-map-adapt.sh: line 10: 29451 Segmentation fault      (core dumped) ./map_adapt -meanfn model/means -varfn model/variances -mixwfn model/mixture_weights -tmatfn model/transition_matrices -accumdir . -mapmeanfn model-adapt/means -mapvarfn model-adapt/variances -mapmixwfn model-adapt/mixture_weights -maptmatfn model-adapt/transition_matrices

Backtrace:

(gdb) bt
#0  0x00000000004030bc in estimate_tau (wt_dcount=<optimized out>, wt_mixw=<optimized out>, 
    wt_mean=<optimized out>, veclen=<optimized out>, n_mixw=<optimized out>, 
    n_density=<optimized out>, n_stream=<optimized out>, n_cb=<optimized out>, 
    si_mixw=<optimized out>, si_var=<optimized out>, si_mean=<optimized out>) at main.c:96
#1  map_update () at main.c:478
#2  main (argc=0, argv=0x2a) at main.c:614

Code:

(gdb) l
91              si_mu = si_mean[0][j][k][m];
92              si_sigma = si_var[0][j][k][m];
93              ml_mu = dnom ? wt_mean[0][j][k][m] / dnom : si_mu;
94              }
95              else { /* Continuous. */
96              dnom = wt_dcount[i][j][k];
97              si_mu = si_mean[i][j][k][m];
98              si_sigma = si_var[i][j][k][m];
99              ml_mu = dnom ? wt_mean[i][j][k][m] / dnom : si_mu;
100             }

Guessing -fixedtau yes may avoid segfault and it does. But I get no improvement in accuracy. (Don't know if I would have otherwise.)

SphinxTrain::Util::TiedWaitForConvergence() is broken in many ways

With large numbers of Queue::POSIX parts, iteration 3 will never run as norm_and_launch_bw.pl gets stuck waiting on zombies. This may be due to some difference between WaitForConvergence() used in iterations 2..N and TiedWaitForConvergence() used in iteration 1, which effectively waits for all of the subsequent iterations to complete.

Also it doesn't correctly retrieve the loss value in step 30, at least not when the script is run manually.

It's pretty much a mess (for which I am entirely responsible, 10 years ago)

`Acoustic model adaptation` tutorial seemingly has not been updated yet

I was readding the article for how to adapt the acoustic model for pocketsphinx and I noticed it saying to install "SphinxBase" wich as been has recently been removed.

Adapting the default acoustic model:

You should also make sure that you have downloaded and compiled sphinxbase and sphinxtrain.

SphinxTrain 5.0.0 is released!:

• The dependency on SphinxBase is gone, because SphinxBase is gone

I feel like this need to be updated so as to not confuse others. (like me)