Menu
▾
▴
kaldi-developers — Kaldi Developers
You can subscribe to this list here.
| 2011 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
(4) |
Jul
|
Aug
|
Sep
(1) |
Oct
(4) |
Nov
(1) |
Dec
(14) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2012 |
Jan
(1) |
Feb
(8) |
Mar
|
Apr
(1) |
May
(3) |
Jun
(13) |
Jul
(7) |
Aug
(11) |
Sep
(6) |
Oct
(14) |
Nov
(16) |
Dec
(1) |
| 2013 |
Jan
(3) |
Feb
(8) |
Mar
(17) |
Apr
(21) |
May
(27) |
Jun
(11) |
Jul
(11) |
Aug
(21) |
Sep
(39) |
Oct
(17) |
Nov
(39) |
Dec
(28) |
| 2014 |
Jan
(36) |
Feb
(30) |
Mar
(35) |
Apr
(17) |
May
(22) |
Jun
(28) |
Jul
(23) |
Aug
(41) |
Sep
(17) |
Oct
(10) |
Nov
(22) |
Dec
(56) |
| 2015 |
Jan
(30) |
Feb
(32) |
Mar
(37) |
Apr
(28) |
May
(79) |
Jun
(18) |
Jul
(35) |
Aug
|
Sep
(1) |
Oct
|
Nov
|
Dec
|
|
From: Patrick P. <pat...@gm...> - 2014-05-21 08:37:04
|
Hi Karel, thanks for your reply. I've checked and finally found my mistake. I've taken the usage "Usage: $0 <data-train> *<data-dev>* <lang-dir> <ali-train> <ali-dev> <exp-dir>" too literally and provided the data of my development set as "<data-dev>". The usage description might be a bit confusing at this point, but luckily I figured it out. I've got one other question: I've noticed that there's only 1 CPU core used while training with a GPU. Since it's always at 100% utilization, wouldn't it make sense to use the other available cores as well? Is there a multithreading option? Thank you very much. Regards, Patrick 2014-05-20 17:25 GMT+02:00 Vesely Karel <ive...@fi...>: > Hard to say what went wrong. Can you double check that the alignments > for the corresponding utterances were correctly generated? > (for example by ali-to-pdf with ascii output and grepping for specific key) > If no targets are found, none of the frames get scored for cross-entropy. > K. > > > > On 05/20/2014 04:47 PM, Daniel Povey wrote: > > I can't answer this offhand as I'm not familiar with Karel's scripts. > Karel might be able to answer. > Dan > > > > On Tue, May 20, 2014 at 10:45 AM, Patrick Proba <pat...@gm...>wrote: > >> Hello Dan, >> >> thanks for your answer. >> You're right, there were only warnings in the main log file. >> I've attached the log file where I found actual errors >> (iter00.initial.log). >> >> The error which appears twice in a row is: >> >>> ERROR (nnet-train-frmshuff:Max():kaldi-vector.cc:515) Empty vector >> >> >> The warnings leading to this seem to be: >> >>> WARNING (nnet-train-frmshuff:main():nnet-train-frmshuff.cc:133) *, >>> missing targets >> >> >> There are 630 IDs listed as missing targets. This is the size of my dev >> set. Either I provided a wrong directory or I've missed a step. >> Can you tell from the log file where the script is expecting data of >> these utterances that's missing? >> My guess would be that it expects to find alignments for the dev set in >> $ali but doesn't find them, is this correct? >> >> Thanks in advance, >> Patrick >> >> >> 2014-05-20 16:23 GMT+02:00 Daniel Povey <dp...@gm...>: >> >> It looks to me like the warnings in the file you attached are not >>> really errors. Most likely there is an error in one of the log files >>> created by train_nnet_scheduler.sh. Look for the most recent file in >>> $dir/log. >>> Dan >>> >>> >>> >>> On Tue, May 20, 2014 at 8:29 AM, Patrick Proba <pat...@gm... >>> > wrote: >>> >>>> Dear Kaldi developers, >>>> >>>> after calling the script "steps/nnet/train.sh" with the following >>>> parameters, I get the message "run.pl: job failed, log is in >>>> exp/dnn5b_pretrain-dbn_dnn/log/train_nnet.log" in the section "# RUNNING >>>> THE NN-TRAINING SCHEDULER". >>>> >>>> "run.pl" $dir/log/train_nnet.log steps/nnet/train.sh >>>> --feature-transform $feature_transform --dbn $dbn --hid-layers 0 >>>> --learn-rate 0.008 data-fmllr-tri4b/train data-fmllr-tri4b/dev data/lang >>>> $ali $ali $dir >>>> >>>> $dir = exp/dnn5b_pretrain-dbn_dnn >>>> $feature_transform = exp/dnn5b_pretrain-dbn/final.feature_transform >>>> $dbn = exp/dnn5b_pretrain-dbn/6.dbn >>>> $ali = exp/tri4b_ali >>>> >>>> I can't figure out why I get this error message nor why it says that >>>> it's manually selecting the CPU in the log file which I've attached. >>>> >>>> System specs: >>>> - Ubuntu 14.04 LTS >>>> - NVIDIA GTX 780, driver 331.62, CUDA V6.0.1 >>>> - Intel Core i5-3570 >>>> - 16GB RAM >>>> >>>> Any help would be greatly appreciated. >>>> >>>> Best regards, >>>> Patrick >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE >>>> Instantly run your Selenium tests across 300+ browser/OS combos. >>>> Get unparalleled scalability from the best Selenium testing platform >>>> available >>>> Simple to use. Nothing to install. Get started now for free." >>>> http://p.sf.net/sfu/SauceLabs >>>> _______________________________________________ >>>> Kaldi-developers mailing list >>>> Kal...@li... >>>> https://lists.sourceforge.net/lists/listinfo/kaldi-developers >>>> >>>> >>> >> > > |
|
From: Vesely K. <ive...@fi...> - 2014-05-20 15:25:43
|
Hard to say what went wrong. Can you double check that the alignments for the corresponding utterances were correctly generated? (for example by ali-to-pdf with ascii output and grepping for specific key) If no targets are found, none of the frames get scored for cross-entropy. K. On 05/20/2014 04:47 PM, Daniel Povey wrote: > I can't answer this offhand as I'm not familiar with Karel's scripts. > Karel might be able to answer. > Dan > > > > On Tue, May 20, 2014 at 10:45 AM, Patrick Proba > <pat...@gm... <mailto:pat...@gm...>> wrote: > > Hello Dan, > > thanks for your answer. > You're right, there were only warnings in the main log file. > I've attached the log file where I found actual errors > (iter00.initial.log). > > The error which appears twice in a row is: > > ERROR (nnet-train-frmshuff:Max():kaldi-vector.cc:515) Empty vector > > > The warnings leading to this seem to be: > > WARNING > (nnet-train-frmshuff:main():nnet-train-frmshuff.cc:133) *, > missing targets > > > There are 630 IDs listed as missing targets. This is the size of > my dev set. Either I provided a wrong directory or I've missed a step. > Can you tell from the log file where the script is expecting data > of these utterances that's missing? > My guess would be that it expects to find alignments for the dev > set in $ali but doesn't find them, is this correct? > > Thanks in advance, > Patrick > > > 2014-05-20 16:23 GMT+02:00 Daniel Povey <dp...@gm... > <mailto:dp...@gm...>>: > > It looks to me like the warnings in the file you attached are > not really errors. Most likely there is an error in one of > the log files created by train_nnet_scheduler.sh. Look for > the most recent file in $dir/log. > Dan > > > > On Tue, May 20, 2014 at 8:29 AM, Patrick Proba > <pat...@gm... <mailto:pat...@gm...>> wrote: > > Dear Kaldi developers, > > after calling the script "steps/nnet/train.sh" with the > following parameters, I get the message "run.pl > <http://run.pl>: job failed, log is in > exp/dnn5b_pretrain-dbn_dnn/log/train_nnet.log" in the > section "# RUNNING THE NN-TRAINING SCHEDULER". > > "run.pl <http://run.pl>" $dir/log/train_nnet.log > steps/nnet/train.sh --feature-transform $feature_transform > --dbn $dbn --hid-layers 0 --learn-rate 0.008 > data-fmllr-tri4b/train data-fmllr-tri4b/dev data/lang $ali > $ali $dir > > $dir = exp/dnn5b_pretrain-dbn_dnn > $feature_transform = > exp/dnn5b_pretrain-dbn/final.feature_transform > $dbn = exp/dnn5b_pretrain-dbn/6.dbn > $ali = exp/tri4b_ali > > I can't figure out why I get this error message nor why it > says that it's manually selecting the CPU in the log file > which I've attached. > > System specs: > - Ubuntu 14.04 LTS > - NVIDIA GTX 780, driver 331.62, CUDA V6.0.1 > - Intel Core i5-3570 > - 16GB RAM > > Any help would be greatly appreciated. > > Best regards, > Patrick > > ------------------------------------------------------------------------------ > "Accelerate Dev Cycles with Automated Cross-Browser > Testing - For FREE > Instantly run your Selenium tests across 300+ browser/OS > combos. > Get unparalleled scalability from the best Selenium > testing platform available > Simple to use. Nothing to install. Get started now for free." > http://p.sf.net/sfu/SauceLabs > _______________________________________________ > Kaldi-developers mailing list > Kal...@li... > <mailto:Kal...@li...> > https://lists.sourceforge.net/lists/listinfo/kaldi-developers > > > > |
|
From: Daniel P. <dp...@gm...> - 2014-05-20 14:47:52
|
I can't answer this offhand as I'm not familiar with Karel's scripts. Karel might be able to answer. Dan On Tue, May 20, 2014 at 10:45 AM, Patrick Proba <pat...@gm...>wrote: > Hello Dan, > > thanks for your answer. > You're right, there were only warnings in the main log file. > I've attached the log file where I found actual errors > (iter00.initial.log). > > The error which appears twice in a row is: > >> ERROR (nnet-train-frmshuff:Max():kaldi-vector.cc:515) Empty vector > > > The warnings leading to this seem to be: > >> WARNING (nnet-train-frmshuff:main():nnet-train-frmshuff.cc:133) *, >> missing targets > > > There are 630 IDs listed as missing targets. This is the size of my dev > set. Either I provided a wrong directory or I've missed a step. > Can you tell from the log file where the script is expecting data of these > utterances that's missing? > My guess would be that it expects to find alignments for the dev set in > $ali but doesn't find them, is this correct? > > Thanks in advance, > Patrick > > > 2014-05-20 16:23 GMT+02:00 Daniel Povey <dp...@gm...>: > > It looks to me like the warnings in the file you attached are not really >> errors. Most likely there is an error in one of the log files created by >> train_nnet_scheduler.sh. Look for the most recent file in $dir/log. >> Dan >> >> >> >> On Tue, May 20, 2014 at 8:29 AM, Patrick Proba <pat...@gm...>wrote: >> >>> Dear Kaldi developers, >>> >>> after calling the script "steps/nnet/train.sh" with the following >>> parameters, I get the message "run.pl: job failed, log is in >>> exp/dnn5b_pretrain-dbn_dnn/log/train_nnet.log" in the section "# RUNNING >>> THE NN-TRAINING SCHEDULER". >>> >>> "run.pl" $dir/log/train_nnet.log steps/nnet/train.sh >>> --feature-transform $feature_transform --dbn $dbn --hid-layers 0 >>> --learn-rate 0.008 data-fmllr-tri4b/train data-fmllr-tri4b/dev data/lang >>> $ali $ali $dir >>> >>> $dir = exp/dnn5b_pretrain-dbn_dnn >>> $feature_transform = exp/dnn5b_pretrain-dbn/final.feature_transform >>> $dbn = exp/dnn5b_pretrain-dbn/6.dbn >>> $ali = exp/tri4b_ali >>> >>> I can't figure out why I get this error message nor why it says that >>> it's manually selecting the CPU in the log file which I've attached. >>> >>> System specs: >>> - Ubuntu 14.04 LTS >>> - NVIDIA GTX 780, driver 331.62, CUDA V6.0.1 >>> - Intel Core i5-3570 >>> - 16GB RAM >>> >>> Any help would be greatly appreciated. >>> >>> Best regards, >>> Patrick >>> >>> >>> ------------------------------------------------------------------------------ >>> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE >>> Instantly run your Selenium tests across 300+ browser/OS combos. >>> Get unparalleled scalability from the best Selenium testing platform >>> available >>> Simple to use. Nothing to install. Get started now for free." >>> http://p.sf.net/sfu/SauceLabs >>> _______________________________________________ >>> Kaldi-developers mailing list >>> Kal...@li... >>> https://lists.sourceforge.net/lists/listinfo/kaldi-developers >>> >>> >> > |
|
From: Daniel P. <dp...@gm...> - 2014-05-20 14:23:10
|
It looks to me like the warnings in the file you attached are not really errors. Most likely there is an error in one of the log files created by train_nnet_scheduler.sh. Look for the most recent file in $dir/log. Dan On Tue, May 20, 2014 at 8:29 AM, Patrick Proba <pat...@gm...>wrote: > Dear Kaldi developers, > > after calling the script "steps/nnet/train.sh" with the following > parameters, I get the message "run.pl: job failed, log is in > exp/dnn5b_pretrain-dbn_dnn/log/train_nnet.log" in the section "# RUNNING > THE NN-TRAINING SCHEDULER". > > "run.pl" $dir/log/train_nnet.log steps/nnet/train.sh --feature-transform > $feature_transform --dbn $dbn --hid-layers 0 --learn-rate 0.008 > data-fmllr-tri4b/train data-fmllr-tri4b/dev data/lang $ali $ali $dir > > $dir = exp/dnn5b_pretrain-dbn_dnn > $feature_transform = exp/dnn5b_pretrain-dbn/final.feature_transform > $dbn = exp/dnn5b_pretrain-dbn/6.dbn > $ali = exp/tri4b_ali > > I can't figure out why I get this error message nor why it says that it's > manually selecting the CPU in the log file which I've attached. > > System specs: > - Ubuntu 14.04 LTS > - NVIDIA GTX 780, driver 331.62, CUDA V6.0.1 > - Intel Core i5-3570 > - 16GB RAM > > Any help would be greatly appreciated. > > Best regards, > Patrick > > > ------------------------------------------------------------------------------ > "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE > Instantly run your Selenium tests across 300+ browser/OS combos. > Get unparalleled scalability from the best Selenium testing platform > available > Simple to use. Nothing to install. Get started now for free." > http://p.sf.net/sfu/SauceLabs > _______________________________________________ > Kaldi-developers mailing list > Kal...@li... > https://lists.sourceforge.net/lists/listinfo/kaldi-developers > > |
|
From: Saman M. <smo...@gm...> - 2014-05-19 10:52:06
|
To whom it may concern, Is there any way to change a specific c++ code in Kaldi and then compile it, without needing to compile all the codes in Kaldi from the begining? I want to change a piece of code and recompile only that cc file (not all cc file in Kaldi) Best Saman |
|
From: <jen...@a2...> - 2014-05-14 15:02:50
|
Kaldi - Build # 527 - Fixed: See the build log in attachment for the details. |
|
From: <jen...@a2...> - 2014-05-14 10:52:51
|
Kaldi - Build # 526 - Failure: See the build log in attachment for the details. |
|
From: <jen...@a2...> - 2014-05-01 06:29:22
|
Kaldi - Build # 519 - Still Failing: See the build log in attachment for the details. |
|
From: <jen...@a2...> - 2014-04-26 23:47:05
|
Kaldi - Build # 518 - Failure: See the build log in attachment for the details. |
|
From: Daniel P. <dp...@gm...> - 2014-04-26 17:40:15
|
cc'ing the developers list so the answer is searchable online. It is possible to convert Kaldi lattices to htk format by using the script utils/convert_slf.pl. It is probably a good idea to use lattice-align-words or lattice-align-words-lexicon first, so that the word timing information is correct. Dan On Sat, Apr 26, 2014 at 2:14 AM, Xinglong Gao <gao...@gm...>wrote: > Could I use some tools for converting the kaldi lattice to htk format? I > want to use SRI tools, such as lattice-tool for downstream process.Thank > you very much. > > > xinglong gao > |
|
From: Ho Y. C. <ric...@gm...> - 2014-04-26 09:34:42
|
For anyone who might be interested in this topic. Though the compression is lossy, the performance of the acoustic models trained with Kaldi using compressed features might not worse than non-compressed features. I did some works in the past (on revision r3805) that compare compressed/non-compressed mfcc features on several hundred hours data (up to boosted MMI GMM stage) and models from compressed features out-performed models on non-compressed features (in several % relative error rate). Chan > Message: 1 > Date: Sun, 13 Apr 2014 17:02:57 -0700 > From: Daniel Povey <dp...@gm...> > Subject: Re: [Kaldi-developers] copy-feats --compress > To: Hang Su <suh...@gm...> > Cc: "kal...@li..." > <kal...@li...> > Message-ID: > <CAEWAuySdcYSxrRRwqtgQwdJDOQdYP_OhheFt= > ozP...@ma...> > Content-Type: text/plain; charset="iso-8859-1" > > The compression is lossy, so differences are expected. It's not possible > to significantly compress floating point numbers of this type, so this > should not be too much of a surprise. > Dan > > > > On Sun, Apr 13, 2014 at 11:09 AM, Hang Su <suh...@gm...> wrote: > > > Hi, > > > > I tried to play with copy-feats, and tracked down a mismatch produced by > > it with/without compression: > > > > cat 1.txt | copy-feats ark,t,cs:- 'scp,t,p:echo SPEAKER1_0000394_0000442 > > -|' | head -2 > > and > > cat 1.txt | copy-feats --compress=true ark,t,cs:- 'scp,t,p:echo > > SPEAKER1_0000394_0000442 -|' | head -2 > > > > gives: > > copy-feats ark,t,cs:- 'scp,t,p:echo SPEAKER1_0000394_0000442 -|' > > [ > > 3.30226 4.87654 14.5157 8.71772 12.2793 -0.437386 19.8966 1.88424 > > 2.96513 11.7998 9.9702 5.18083 98.2586 0.704592 -0.552088 > > and > > copy-feats --compress=true ark,t,cs:- 'scp,t,p:echo > > SPEAKER1_0000394_0000442 -|' > > LOG (copy-feats:main():copy-feats.cc:95) Copied 1 feature matrices. > > [ > > 3.276405 4.868782 14.53272 8.780814 12.22998 -0.4491575 19.91133 > > 1.896849 2.975965 11.76886 9.92212 5.19049 98.25037 0.7003881 -0.5524781 > > respectively. > > > > 1.txt is attached. > > > > more info: > > I'm using version 3883. Nevertheless, looking at the svn log, I found the > > most recent update relating to compress is done in r3339. Also, previous > > bugs that relates to compress have something to do with platform, so I > > wonder if this is the case again here. I tested it on ubuntu linux and > > redhat linux, both of them have this issue. In addition, as I remove some > > lines in the matrix, the difference between with/without compression > output > > becomes smaller. > > > > > > Thanks a lot, > > Hang Su > > > > > > > |
|
From: <jen...@a2...> - 2014-04-26 03:37:13
|
Kaldi - Build # 515 - Still Failing: See the build log in attachment for the details. |
|
From: Simon K. <sim...@gm...> - 2014-04-22 21:46:37
|
Mea culpa. Indeed it is this bug. I had changed my OS since I had last encountered it, when you had pointed me to the problem/solution. Since then I had of course forgotten to verify to use the appropriate version of gcc. It works like a charm now, thanks a lot. Simon On 04/22/2014 06:37 PM, Daniel Povey wrote: > I think this is a bug in gcc version 4.8.1 that causes random crashes in > "nth_element". > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58800 > This can't easily be worked around since nth_element is used in many > places. But gcc has since been fixed. > Dan > > > > On Tue, Apr 22, 2014 at 10:30 AM, Simon Klüpfel > <sim...@gm... <mailto:sim...@gm...>> wrote: > > Hi, > > I am experiencing yet another problem using the online decoder from > the sandbox. I am using revision 3940. > > Everything up to the decoding stage works fine, and here it seems to > throw segfaults rather randomly. while I managed to have a > successful decoding run (see output attached), in most cases it just > dies at some point. > > I attached the valgrind output. It shows once again 'Conditional > jump or move depends on uninitialised value(s)', but now in the end > also states that "the 'impossible' happened". > > I am not sure if there is another bug in the code, or if my system > might be faulty. > > I am using MFCCs without transforms. > > Thanks, > > Simon > > > > On 04/14/2014 09:15 PM, Daniel Povey wrote: > > > Oh-- I see. I fixed it, thanks. > Dan > > > > On Mon, Apr 14, 2014 at 4:57 PM, Simon Klüpfel > <sim...@gm... <mailto:sim...@gm...> > <mailto:simon.kluepfel@gmail.__com > <mailto:sim...@gm...>>> wrote: > > Hi Dan, > > did you also have a look at the valgrind output? This is a > different > issue, I believe. > > However, I am not sure if this is a bug in the C++ code, or a > problem with my data or setup. > > I had used the higher frequency for both training and testing > > All the best, > > Simon > > > On 04/14/2014 07:21 PM, Daniel Povey wrote: > > Thanks for the bug report-- fixed it. > Regarding setting --sample-frequency=41000 in > conf/mfcc.conf, > this is OK > as long as you do so in both train and test conditions, > since > features > (as you probably know) are not comparable across > different sample > frequencies. > Dan > > > > On Mon, Apr 14, 2014 at 8:49 AM, Simon Klüpfel > <sim...@gm... > <mailto:sim...@gm...> > <mailto:simon.kluepfel@gmail.__com > <mailto:sim...@gm...>> > <mailto:simon.kluepfel@gmail. > <mailto:simon.kluepfel@gmail.>____com > > <mailto:simon.kluepfel@gmail.__com > <mailto:sim...@gm...>>>> wrote: > > Hi, > > there is a small bug in the prepare_online_decoding.sh > script, I > attached the diff. > > I also experience a segfault. I am not sure if > this is due > to the > data I am using, or a bug. I attached the output > of the > command, as > well as the valgrind output reporting use of several > uninitialized > values. > > My recordings are 441kHz, so I added > --sample-frequency=44100 > to conf/mfcc.conf before running > run_online_decoding.sh > > Please tell me if you need any more info on my > setup or > anything else. > > Simon > > > > ------------------------------____----------------------------__--__------------------ > > Learn Graph Databases - Download FREE O'Reilly Book > "Graph Databases" is the definitive new guide to graph > databases and > their > applications. Written by three acclaimed leaders > in the field, > this first edition is now available. Download your > free > book today! > http://p.sf.net/sfu/NeoTech > ___________________________________________________ > Kaldi-developers mailing list > Kaldi-developers@lists.__sourc__eforge.net > <http://sourceforge.net> > <mailto:Kaldi-developers@__lists.sourceforge.net > <mailto:Kal...@li...>> > <mailto:Kaldi-developers@ > <mailto:Kaldi-developers@>__lis__ts.sourceforge.net > <http://lists.sourceforge.net> > <mailto:Kaldi-developers@__lists.sourceforge.net > <mailto:Kal...@li...>>> > https://lists.sourceforge.net/____lists/listinfo/kaldi-____developers > <https://lists.sourceforge.net/__lists/listinfo/kaldi-__developers> > <https://lists.sourceforge.__net/lists/listinfo/kaldi-__developers > <https://lists.sourceforge.net/lists/listinfo/kaldi-developers>> > > > > > > |
|
From: Daniel P. <dp...@gm...> - 2014-04-22 18:37:40
|
I think this is a bug in gcc version 4.8.1 that causes random crashes in "nth_element". http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58800 This can't easily be worked around since nth_element is used in many places. But gcc has since been fixed. Dan On Tue, Apr 22, 2014 at 10:30 AM, Simon Klüpfel <sim...@gm...>wrote: > Hi, > > I am experiencing yet another problem using the online decoder from the > sandbox. I am using revision 3940. > > Everything up to the decoding stage works fine, and here it seems to throw > segfaults rather randomly. while I managed to have a successful decoding > run (see output attached), in most cases it just dies at some point. > > I attached the valgrind output. It shows once again 'Conditional jump or > move depends on uninitialised value(s)', but now in the end also states > that "the 'impossible' happened". > > I am not sure if there is another bug in the code, or if my system might > be faulty. > > I am using MFCCs without transforms. > > Thanks, > > Simon > > > > On 04/14/2014 09:15 PM, Daniel Povey wrote: > >> >> Oh-- I see. I fixed it, thanks. >> Dan >> >> >> >> On Mon, Apr 14, 2014 at 4:57 PM, Simon Klüpfel <sim...@gm... >> <mailto:sim...@gm...>> wrote: >> >> Hi Dan, >> >> did you also have a look at the valgrind output? This is a different >> issue, I believe. >> >> However, I am not sure if this is a bug in the C++ code, or a >> problem with my data or setup. >> >> I had used the higher frequency for both training and testing >> >> All the best, >> >> Simon >> >> >> On 04/14/2014 07:21 PM, Daniel Povey wrote: >> >> Thanks for the bug report-- fixed it. >> Regarding setting --sample-frequency=41000 in conf/mfcc.conf, >> this is OK >> as long as you do so in both train and test conditions, since >> features >> (as you probably know) are not comparable across different sample >> frequencies. >> Dan >> >> >> >> On Mon, Apr 14, 2014 at 8:49 AM, Simon Klüpfel >> <sim...@gm... <mailto:sim...@gm...> >> <mailto:simon.kluepfel@gmail.__com >> >> <mailto:sim...@gm...>>> wrote: >> >> Hi, >> >> there is a small bug in the prepare_online_decoding.sh >> script, I >> attached the diff. >> >> I also experience a segfault. I am not sure if this is due >> to the >> data I am using, or a bug. I attached the output of the >> command, as >> well as the valgrind output reporting use of several >> uninitialized >> values. >> >> My recordings are 441kHz, so I added >> --sample-frequency=44100 >> to conf/mfcc.conf before running run_online_decoding.sh >> >> Please tell me if you need any more info on my setup or >> anything else. >> >> Simon >> >> >> ------------------------------__---------------------------- >> --__------------------ >> >> Learn Graph Databases - Download FREE O'Reilly Book >> "Graph Databases" is the definitive new guide to graph >> databases and >> their >> applications. Written by three acclaimed leaders in the >> field, >> this first edition is now available. Download your free >> book today! >> http://p.sf.net/sfu/NeoTech >> _________________________________________________ >> Kaldi-developers mailing list >> Kaldi-developers@lists.__sourceforge.net >> <mailto:Kal...@li...> >> <mailto:Kaldi-developers@__lists.sourceforge.net >> <mailto:Kal...@li...>> >> https://lists.sourceforge.net/__lists/listinfo/kaldi-__developers< >> https://lists.sourceforge.net/lists/listinfo/kaldi-developers> >> >> >> >> >> > |
|
From: Simon K. <sim...@gm...> - 2014-04-22 14:31:09
|
Hi, I am experiencing yet another problem using the online decoder from the sandbox. I am using revision 3940. Everything up to the decoding stage works fine, and here it seems to throw segfaults rather randomly. while I managed to have a successful decoding run (see output attached), in most cases it just dies at some point. I attached the valgrind output. It shows once again 'Conditional jump or move depends on uninitialised value(s)', but now in the end also states that "the 'impossible' happened". I am not sure if there is another bug in the code, or if my system might be faulty. I am using MFCCs without transforms. Thanks, Simon On 04/14/2014 09:15 PM, Daniel Povey wrote: > > Oh-- I see. I fixed it, thanks. > Dan > > > > On Mon, Apr 14, 2014 at 4:57 PM, Simon Klüpfel <sim...@gm... > <mailto:sim...@gm...>> wrote: > > Hi Dan, > > did you also have a look at the valgrind output? This is a different > issue, I believe. > > However, I am not sure if this is a bug in the C++ code, or a > problem with my data or setup. > > I had used the higher frequency for both training and testing > > All the best, > > Simon > > > On 04/14/2014 07:21 PM, Daniel Povey wrote: > > Thanks for the bug report-- fixed it. > Regarding setting --sample-frequency=41000 in conf/mfcc.conf, > this is OK > as long as you do so in both train and test conditions, since > features > (as you probably know) are not comparable across different sample > frequencies. > Dan > > > > On Mon, Apr 14, 2014 at 8:49 AM, Simon Klüpfel > <sim...@gm... <mailto:sim...@gm...> > <mailto:simon.kluepfel@gmail.__com > <mailto:sim...@gm...>>> wrote: > > Hi, > > there is a small bug in the prepare_online_decoding.sh > script, I > attached the diff. > > I also experience a segfault. I am not sure if this is due > to the > data I am using, or a bug. I attached the output of the > command, as > well as the valgrind output reporting use of several > uninitialized > values. > > My recordings are 441kHz, so I added > --sample-frequency=44100 > to conf/mfcc.conf before running run_online_decoding.sh > > Please tell me if you need any more info on my setup or > anything else. > > Simon > > > ------------------------------__------------------------------__------------------ > Learn Graph Databases - Download FREE O'Reilly Book > "Graph Databases" is the definitive new guide to graph > databases and > their > applications. Written by three acclaimed leaders in the field, > this first edition is now available. Download your free > book today! > http://p.sf.net/sfu/NeoTech > _________________________________________________ > Kaldi-developers mailing list > Kaldi-developers@lists.__sourceforge.net > <mailto:Kal...@li...> > <mailto:Kaldi-developers@__lists.sourceforge.net > <mailto:Kal...@li...>> > https://lists.sourceforge.net/__lists/listinfo/kaldi-__developers <https://lists.sourceforge.net/lists/listinfo/kaldi-developers> > > > > |
|
From: Daniel P. <dp...@gm...> - 2014-04-14 21:16:00
|
Oh-- I see. I fixed it, thanks. Dan On Mon, Apr 14, 2014 at 4:57 PM, Simon Klüpfel <sim...@gm...>wrote: > Hi Dan, > > did you also have a look at the valgrind output? This is a different > issue, I believe. > > However, I am not sure if this is a bug in the C++ code, or a problem with > my data or setup. > > I had used the higher frequency for both training and testing > > All the best, > > Simon > > > On 04/14/2014 07:21 PM, Daniel Povey wrote: > >> Thanks for the bug report-- fixed it. >> Regarding setting --sample-frequency=41000 in conf/mfcc.conf, this is OK >> as long as you do so in both train and test conditions, since features >> (as you probably know) are not comparable across different sample >> frequencies. >> Dan >> >> >> >> On Mon, Apr 14, 2014 at 8:49 AM, Simon Klüpfel <sim...@gm... >> <mailto:sim...@gm...>> wrote: >> >> Hi, >> >> there is a small bug in the prepare_online_decoding.sh script, I >> attached the diff. >> >> I also experience a segfault. I am not sure if this is due to the >> data I am using, or a bug. I attached the output of the command, as >> well as the valgrind output reporting use of several uninitialized >> values. >> >> My recordings are 441kHz, so I added >> --sample-frequency=44100 >> to conf/mfcc.conf before running run_online_decoding.sh >> >> Please tell me if you need any more info on my setup or anything else. >> >> Simon >> >> ------------------------------------------------------------ >> ------------------ >> Learn Graph Databases - Download FREE O'Reilly Book >> "Graph Databases" is the definitive new guide to graph databases and >> their >> applications. Written by three acclaimed leaders in the field, >> this first edition is now available. Download your free book today! >> http://p.sf.net/sfu/NeoTech >> _______________________________________________ >> Kaldi-developers mailing list >> Kal...@li... >> <mailto:Kal...@li...> >> https://lists.sourceforge.net/lists/listinfo/kaldi-developers >> >> >> > |
|
From: Simon K. <sim...@gm...> - 2014-04-14 20:57:15
|
Hi Dan, did you also have a look at the valgrind output? This is a different issue, I believe. However, I am not sure if this is a bug in the C++ code, or a problem with my data or setup. I had used the higher frequency for both training and testing All the best, Simon On 04/14/2014 07:21 PM, Daniel Povey wrote: > Thanks for the bug report-- fixed it. > Regarding setting --sample-frequency=41000 in conf/mfcc.conf, this is OK > as long as you do so in both train and test conditions, since features > (as you probably know) are not comparable across different sample > frequencies. > Dan > > > > On Mon, Apr 14, 2014 at 8:49 AM, Simon Klüpfel <sim...@gm... > <mailto:sim...@gm...>> wrote: > > Hi, > > there is a small bug in the prepare_online_decoding.sh script, I > attached the diff. > > I also experience a segfault. I am not sure if this is due to the > data I am using, or a bug. I attached the output of the command, as > well as the valgrind output reporting use of several uninitialized > values. > > My recordings are 441kHz, so I added > --sample-frequency=44100 > to conf/mfcc.conf before running run_online_decoding.sh > > Please tell me if you need any more info on my setup or anything else. > > Simon > > ------------------------------------------------------------------------------ > Learn Graph Databases - Download FREE O'Reilly Book > "Graph Databases" is the definitive new guide to graph databases and > their > applications. Written by three acclaimed leaders in the field, > this first edition is now available. Download your free book today! > http://p.sf.net/sfu/NeoTech > _______________________________________________ > Kaldi-developers mailing list > Kal...@li... > <mailto:Kal...@li...> > https://lists.sourceforge.net/lists/listinfo/kaldi-developers > > |
|
From: Daniel P. <dp...@gm...> - 2014-04-14 19:21:25
|
Thanks for the bug report-- fixed it. Regarding setting --sample-frequency=41000 in conf/mfcc.conf, this is OK as long as you do so in both train and test conditions, since features (as you probably know) are not comparable across different sample frequencies. Dan On Mon, Apr 14, 2014 at 8:49 AM, Simon Klüpfel <sim...@gm...>wrote: > Hi, > > there is a small bug in the prepare_online_decoding.sh script, I attached > the diff. > > I also experience a segfault. I am not sure if this is due to the data I > am using, or a bug. I attached the output of the command, as well as the > valgrind output reporting use of several uninitialized values. > > My recordings are 441kHz, so I added > --sample-frequency=44100 > to conf/mfcc.conf before running run_online_decoding.sh > > Please tell me if you need any more info on my setup or anything else. > > Simon > > > ------------------------------------------------------------------------------ > Learn Graph Databases - Download FREE O'Reilly Book > "Graph Databases" is the definitive new guide to graph databases and their > applications. Written by three acclaimed leaders in the field, > this first edition is now available. Download your free book today! > http://p.sf.net/sfu/NeoTech > _______________________________________________ > Kaldi-developers mailing list > Kal...@li... > https://lists.sourceforge.net/lists/listinfo/kaldi-developers > > |
|
From: Simon K. <sim...@gm...> - 2014-04-14 12:49:36
|
Hi, there is a small bug in the prepare_online_decoding.sh script, I attached the diff. I also experience a segfault. I am not sure if this is due to the data I am using, or a bug. I attached the output of the command, as well as the valgrind output reporting use of several uninitialized values. My recordings are 441kHz, so I added --sample-frequency=44100 to conf/mfcc.conf before running run_online_decoding.sh Please tell me if you need any more info on my setup or anything else. Simon |
|
From: Simon K. <sim...@gm...> - 2014-04-14 10:54:28
|
Hi, thanks for the explanations. I found the mistake regarding the wrong config file. When adjusting the main script, I passed the original directory to the decoder, not the one created by prepare_online_decoding.sh. Then I probably tried to fix the (to be expected) errors by copying/creating files. Once I fixed this, it did work. All the best, Simon On 04/11/2014 06:56 PM, Daniel Povey wrote: > It's good that you're testing this, > > I think what is going on is that you are somehow mixing up different > types of config in the conf/ subdirectory: the "*.conf" files, which are > intended for use by Kaldi command-line programs, and which have lines like > --X=Y > and the "*.config" files, which are intended to be sourced by bash > scripts, and which have lines like > X=Y > I'll modify that assert statement to be a bit more user friendly. > > The file "config" that you attached (and which I perhaps should have > named "conf", according to that naming scheme), does not look like the > "config" file that is created by "prepare_online_decoding.sh". It > should have lines like > --mfcc-config=exp/tri1_150_20000/conf/mfcc.conf > but the file you attached has lines like > beam=11.0 # beam for decoding. Was 13.0 in the scripts. > first_beam=8.0 # beam for 1st-pass decoding in SAT. > > It looks like you copied it from the conf/ subdirectory > to exp/tri_150_20000/conf/. > > Dan > > > > On Fri, Apr 11, 2014 at 5:02 AM, Simon Klüpfel <sim...@gm... > <mailto:sim...@gm...>> wrote: > > Hi, > > I tried to test the online decoder in sandbox/online, and get an error. > > I used my own models trained earlier, and the > steps/online/prepare_online___decoding.sh step went fine, after I > removed the check for the file 'trans.1' (as I do not have it). If I > got it right from the script, it is not needed in general. > > When trying to decode, I get the error (see attached log file): > > KALDI_ASSERT: at > online2-wav-gmm-latgen-faster:__SplitLongArg:parse-options.cc:__491, > failed: in.substr(0, 2) == "--" > > It seems to me it has problems parsing the options. > > I also attached the config file specified in the call to > online2-wav-gmm-latgen-faster > > I hope you can point me to what is going wrong. > > I'm using revision 3870. > > Thanks a lot, > > Simon > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Kaldi-developers mailing list > Kal...@li... > <mailto:Kal...@li...> > https://lists.sourceforge.net/lists/listinfo/kaldi-developers > > |
|
From: Daniel P. <dp...@gm...> - 2014-04-14 00:03:06
|
The compression is lossy, so differences are expected. It's not possible to significantly compress floating point numbers of this type, so this should not be too much of a surprise. Dan On Sun, Apr 13, 2014 at 11:09 AM, Hang Su <suh...@gm...> wrote: > Hi, > > I tried to play with copy-feats, and tracked down a mismatch produced by > it with/without compression: > > cat 1.txt | copy-feats ark,t,cs:- 'scp,t,p:echo SPEAKER1_0000394_0000442 > -|' | head -2 > and > cat 1.txt | copy-feats --compress=true ark,t,cs:- 'scp,t,p:echo > SPEAKER1_0000394_0000442 -|' | head -2 > > gives: > copy-feats ark,t,cs:- 'scp,t,p:echo SPEAKER1_0000394_0000442 -|' > [ > 3.30226 4.87654 14.5157 8.71772 12.2793 -0.437386 19.8966 1.88424 > 2.96513 11.7998 9.9702 5.18083 98.2586 0.704592 -0.552088 > and > copy-feats --compress=true ark,t,cs:- 'scp,t,p:echo > SPEAKER1_0000394_0000442 -|' > LOG (copy-feats:main():copy-feats.cc:95) Copied 1 feature matrices. > [ > 3.276405 4.868782 14.53272 8.780814 12.22998 -0.4491575 19.91133 > 1.896849 2.975965 11.76886 9.92212 5.19049 98.25037 0.7003881 -0.5524781 > respectively. > > 1.txt is attached. > > more info: > I'm using version 3883. Nevertheless, looking at the svn log, I found the > most recent update relating to compress is done in r3339. Also, previous > bugs that relates to compress have something to do with platform, so I > wonder if this is the case again here. I tested it on ubuntu linux and > redhat linux, both of them have this issue. In addition, as I remove some > lines in the matrix, the difference between with/without compression output > becomes smaller. > > > Thanks a lot, > Hang Su > > > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Kaldi-developers mailing list > Kal...@li... > https://lists.sourceforge.net/lists/listinfo/kaldi-developers > > |
|
From: Hang Su <suh...@gm...> - 2014-04-13 18:10:14
|
SPEAKER1_0000394_0000442 [ 3.30226 4.87654 14.5157 8.71772 12.2793 -0.437386 19.8966 1.88424 2.96513 11.7998 9.9702 5.18083 98.2586 0.704592 -0.552088 2.23857 2.82966 16.829 10.2128 22.1307 21.0935 17.1857 6.15127 5.70233 19.3512 -2.37598 2.96848 97.1666 0.707015 -0.452858 0.245604 6.61683 17.854 17.1648 28.3889 29.4127 18.4919 7.55883 5.11777 -4.17093 -3.33021 1.3005 94.7461 0.704233 -0.286433 0.977615 4.7913 9.25463 3.74941 20.8163 1.32655 16.9175 2.59761 8.12803 3.24004 1.26482 5.29867 93.6731 0.694274 -0.242495 3.54497 3.57513 11.551 7.76833 14.9973 12.6748 14.9907 -2.81681 2.10089 16.0378 -1.95338 10.7706 95.76 0.694543 -0.339899 3.86601 3.09172 8.42541 9.3024 12.1093 7.74815 23.593 3.18749 -1.30701 -9.06865 -10.7393 -2.24292 96.8777 0.745059 -0.260475 2.89999 4.11452 11.1752 8.32794 22.2215 12.8134 22.2947 16.6912 3.27796 6.23366 2.16705 -2.91177 96.9569 0.785615 -0.182026 6.76215 0.838756 15.1109 7.90893 22.5336 16.0641 17.5888 -2.39119 0.422787 0.619726 4.28512 -4.63286 97.4309 0.79387 -0.158819 4.61156 -1.33406 13.1982 2.74434 25.4513 6.28255 14.3637 -0.499257 -4.05216 6.22208 0.0104085 -6.28922 94.7249 0.761838 -0.0477109 5.77292 -1.23484 9.78006 5.17629 22.2743 5.19516 10.4895 -5.47247 13.5096 4.55671 5.96046 3.51451 95.5425 0.814597 -0.110605 4.29052 -0.094997 12.5085 6.50015 11.6271 -1.57857 9.13881 -13.7943 -0.100793 -1.15127 -7.74411 2.09382 93.904 0.823929 -0.0632656 3.89599 -2.97513 14.1939 0.0799069 18.1247 10.0778 4.21545 0.409541 15.2739 0.655898 -3.47107 -1.07119 95.1131 0.79082 -0.120408 5.53214 -3.47762 24.1007 -3.45318 8.84969 13.3562 13.4273 6.72197 4.51334 4.07779 -2.40785 0.731226 94.2899 0.741201 -0.133362 4.86395 -7.62735 10.6633 1.08638 9.24434 15.7185 26.8229 7.67632 15.9168 -7.06326 -6.70101 13.383 95.3261 0.70522 -0.152592 5.21883 0.869287 8.11712 5.72059 9.68357 11.9787 25.4484 1.01501 15.8995 4.00979 0.446432 4.47049 99.036 0.679918 -0.219088 6.93427 -1.19286 9.84082 0.292894 14.5713 3.98877 17.5682 -14.2187 -4.76529 -3.41999 -6.78653 3.67882 98.2039 0.684135 -0.216687 7.70013 5.08135 11.2742 -3.09124 12.9134 7.03899 22.5693 -7.63536 4.91099 -3.99007 -13.2699 0.233322 98.8085 0.712847 -0.24447 5.27975 1.04739 11.2877 -6.56446 4.64892 -1.06015 14.597 -2.60939 13.4341 0.197234 -7.53113 7.18572 98.4727 0.720833 -0.268052 6.62097 0.441847 15.0727 6.95675 14.0759 1.35753 12.4544 3.33736 13.837 16.7048 -10.0416 -4.23288 96.1995 0.726127 -0.187978 8.08306 4.38676 10.3213 8.8625 8.78364 8.89748 15.3843 7.15718 12.0542 3.64373 10.215 3.18257 98.5832 0.69876 -0.108379 11.5439 4.12088 7.2969 0.862252 2.07133 -8.91474 10.3371 -12.3136 1.7377 6.24379 4.87878 1.67392 100.789 0.734112 -0.166622 8.28612 4.62338 16.1337 -0.524253 -2.41015 10.5489 22.5471 9.21576 2.01606 8.53277 1.67904 8.86532 97.8637 0.737343 -0.302163 11.9075 -0.990586 4.53584 -14.3782 0.458059 2.46285 23.1248 13.7322 13.0775 -6.35862 -1.80748 18.6757 103.352 0.718679 -0.384912 10.3922 -13.8532 -4.81511 -6.13013 -3.74932 -9.6239 23.0994 -0.141972 -4.07735 0.375201 -1.25238 4.14519 104.002 0.69535 -0.382837 4.30309 -10.3345 2.49935 -9.88317 -5.14957 -3.261 18.2078 -2.62618 5.86404 15.0785 1.07867 3.84977 107.408 0.695799 -0.408069 3.02667 -2.8759 4.80137 -7.41084 -0.517827 5.30276 31.7145 20.7532 2.14066 1.70489 -8.26902 1.27394 112.451 0.712129 -0.404468 10.4609 -8.21254 -4.0984 -6.6285 -6.90815 -2.29099 36.3982 6.42942 -7.39115 7.10324 -6.22976 4.65305 108.475 0.733484 -0.441079 15.7687 -12.0531 -1.64787 -16.4245 -0.689556 -10.3836 31.1606 4.54949 -6.85299 4.74192 -2.49505 10.7291 106.116 0.715001 -0.22584 14.3433 -5.04363 1.10194 -5.76402 11.5214 0.0712546 15.689 9.08988 3.52318 0.124889 -16.0956 5.52106 105.51 0.527382 -0.199357 16.6216 -3.03365 2.4296 1.01538 5.24172 -7.32523 26.1007 6.84546 11.4736 15.5429 -23.3957 -3.68518 106.49 0.50504 -0.187903 14.0852 -0.427026 10.5407 -0.606385 5.62481 -4.72334 7.9532 18.5975 21.1234 13.2569 -13.1508 11.8079 102.381 0.470765 -0.180901 14.9342 5.3714 2.39359 -15.2733 -4.56338 4.73869 20.9139 13.1627 2.79016 11.3484 -5.37281 3.10457 100.453 0.0078649 -0.144565 17.5973 0.477467 8.13738 -2.46063 -16.0544 -2.06765 13.7971 0.775218 -1.15325 9.66279 7.97288 10.8354 100.808 -0.0608657 -0.0724433 13.0708 1.57278 8.92834 -4.54456 -7.51911 -4.25872 15.1319 10.6497 8.80404 16.3141 3.31077 -0.0869963 100.18 -0.0920008 -0.101252 11.9481 -1.58849 8.02374 -2.75714 8.44679 4.78597 12.8893 0.476681 -5.52614 0.823737 -0.12543 13.4859 99.4811 -0.112907 -0.129736 11.0691 0.390962 6.97624 -0.195724 9.37809 13.1753 17.8126 -13.5437 -5.72497 0.307198 -1.14673 8.05208 99.8358 -0.169255 -0.0759194 6.61613 0.32481 10.4293 7.73214 10.6165 4.93758 13.7035 2.23193 9.52115 2.17513 -1.71189 -13.1603 96.0255 -0.447498 -0.0929995 11.2654 -0.101358 4.35132 -8.62612 1.4422 10.9239 11.8862 -7.29845 2.6775 11.2081 12.8044 6.08038 95.9184 -0.46661 -0.22709 13.2913 4.59667 11.2911 -3.61048 11.9028 2.36992 7.17391 5.83115 18.6526 6.15119 1.63544 -2.92505 95.876 -0.514883 -0.263801 11.6716 4.8371 10.2751 -2.12514 8.09012 -4.0207 8.37221 9.43877 9.93736 -3.7629 2.71376 12.3407 95.5158 -0.50259 -0.502148 7.17215 2.93906 9.44928 2.57311 1.42734 10.2751 16.7222 11.2324 21.028 9.90731 -2.55207 5.29203 97.016 -0.484375 -0.344325 10.2056 4.23029 1.31684 7.80313 11.353 4.53002 15.6715 -0.709072 16.7293 5.78946 4.3522 6.77744 96.6702 -0.46643 -0.318542 5.83674 -3.17613 6.5847 11.6508 14.175 -6.09928 23.4407 -1.50996 22.4423 0.913444 8.01145 -5.48594 92.6491 -0.455573 -0.332522 11.0353 4.92742 3.3432 -0.50894 21.1729 10.2718 12.9702 0.407143 11.6526 1.97401 12.3381 -3.15077 94.2821 -0.448664 -0.283407 4.86685 4.11707 10.7814 7.64304 21.7988 24.1176 13.3083 -5.29862 1.46597 -17.5995 -4.72716 6.60152 93.9229 -0.437179 -0.19438 7.30463 1.84248 17.2892 9.03373 22.8242 19.067 13.094 -2.39119 13.055 -11.5559 3.45835 0.332903 97.1922 -0.429194 -0.315517 5.92764 4.02929 13.1802 10.6986 24.5365 24.4909 7.23105 -4.88019 15.6623 6.32192 -4.69865 5.70197 95.0495 -0.42722 -0.17745 4.95194 0.715358 12.801 5.55493 26.0705 12.5411 8.58807 2.98846 1.86098 -3.10746 -11.6448 1.47144 94.3713 -0.425784 -0.298962 6.62 2.66174 16.2068 7.06533 21.6386 8.52252 13.4273 4.10468 2.59531 -8.13251 3.44997 16.2011 95.8225 -0.422644 -0.215511 6.94491 -0.528798 8.05412 11.7831 12.8952 15.1202 6.02005 -2.91512 6.55331 -7.99072 -0.784497 -0.526812 93.8649 -0.414927 -0.264676 ] |
|
From: Daniel P. <dp...@gm...> - 2014-04-11 18:56:14
|
It's good that you're testing this, I think what is going on is that you are somehow mixing up different types of config in the conf/ subdirectory: the "*.conf" files, which are intended for use by Kaldi command-line programs, and which have lines like --X=Y and the "*.config" files, which are intended to be sourced by bash scripts, and which have lines like X=Y I'll modify that assert statement to be a bit more user friendly. The file "config" that you attached (and which I perhaps should have named "conf", according to that naming scheme), does not look like the "config" file that is created by "prepare_online_decoding.sh". It should have lines like --mfcc-config=exp/tri1_150_20000/conf/mfcc.conf but the file you attached has lines like beam=11.0 # beam for decoding. Was 13.0 in the scripts. first_beam=8.0 # beam for 1st-pass decoding in SAT. It looks like you copied it from the conf/ subdirectory to exp/tri_150_20000/conf/. Dan On Fri, Apr 11, 2014 at 5:02 AM, Simon Klüpfel <sim...@gm...>wrote: > Hi, > > I tried to test the online decoder in sandbox/online, and get an error. > > I used my own models trained earlier, and the steps/online/prepare_online_decoding.sh > step went fine, after I removed the check for the file 'trans.1' (as I do > not have it). If I got it right from the script, it is not needed in > general. > > When trying to decode, I get the error (see attached log file): > > KALDI_ASSERT: at online2-wav-gmm-latgen-faster: > SplitLongArg:parse-options.cc:491, failed: in.substr(0, 2) == "--" > > It seems to me it has problems parsing the options. > > I also attached the config file specified in the call to > online2-wav-gmm-latgen-faster > > I hope you can point me to what is going wrong. > > I'm using revision 3870. > > Thanks a lot, > > Simon > > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Kaldi-developers mailing list > Kal...@li... > https://lists.sourceforge.net/lists/listinfo/kaldi-developers > > |
|
From: Simon K. <sim...@gm...> - 2014-04-11 12:03:08
|
Hi, I tried to test the online decoder in sandbox/online, and get an error. I used my own models trained earlier, and the steps/online/prepare_online_decoding.sh step went fine, after I removed the check for the file 'trans.1' (as I do not have it). If I got it right from the script, it is not needed in general. When trying to decode, I get the error (see attached log file): KALDI_ASSERT: at online2-wav-gmm-latgen-faster:SplitLongArg:parse-options.cc:491, failed: in.substr(0, 2) == "--" It seems to me it has problems parsing the options. I also attached the config file specified in the call to online2-wav-gmm-latgen-faster I hope you can point me to what is going wrong. I'm using revision 3870. Thanks a lot, Simon |
|
From: <jen...@a2...> - 2014-04-01 08:02:53
|
Kaldi - Build # 511 - Still Failing: See the build log in attachment for the details. |
23 messages has been excluded from this view by a project administrator.
