3) Next step was to run "python ..\sphinxtrain\scripts\sphinxtrain run" but that does not work for me so I tried "python "D:/sphinx/sphinxtrain/scripts/sphinxtrain" run" and miraculously it run but an error was showing up. I attached below the screenshot of my failed training. I checked cmusphinx for this error and it says
"The .mfc files are the feature files converted from the input audio files on stage 000.comp_feats. Did you skip this step? Did you add new audio files without converting them? The training process expects a feature file to be there, and it isn't."
Thanks for patiently replying to my questions sir.
1) I followed what you said, I placed my database folder outside the sphinxtrain folder and now they are in one single folder. (For now I am using CMU's AN4 database, I will change this to my files later on when training the files are already succesful, I want to master the steps first)
2) I run "python ../sphinxtrain/scripts/sphinxtrain run"
3) It said configuration (sphinx_train.cfg) not defined but I am pretty sure I already configured it. But just to make it sure, I attached here my sphinx_train.cfg file.
4) I attached below the picture of the results when I run the line at #2.
Hoping to hear from you again since I am really illiterate when it comes to running the command prompt. Its my first time using it.
1) I redownload Sphinxtrain and AN4(So there would be no corrupt files), placed them inside one folder. Run the script "python ../sphinxtrain/scripts/sphinxtrain -t an4 setup"
2) Setting the database folder was successful, I then configured the sphinxtrain\etc\sphinx_train.cfg file.
This is wrong because you need to edit an4/etc/sphinx_train.cfg, not sphinxtrain/etc/sphinx_train.cfg
I suspect you also made many other mistakes. So it is hard to help you. I suggest you to find someone more experienced than you locally so he can help you.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Even if I replaced everything with new files. New sphinxtrain and new an4. Configured the one in an4\etc\sphinx_train.cgf (By the way, the CMU tutorial did not mention it was the file from an4, because there was also an etc file in the sphinxtrain folder so I just followed orders) the .MFC files are still not generating.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello Leimiaoren,
I had same problem and after this setup:
"python ../sphinxtrain/scripts/sphinxtrain -t an4 setup"
I only changed these lines and everything is ok:
$CFG_WAVFILES_DIR = "$CFG_BASE_DIR/wav";
$CFG_WAVFILE_EXTENSION = 'sph';
$CFG_WAVFILE_TYPE = 'nist'; # one of nist, mswav, raw
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I have encountered the same problem as you.
This problem has been bothering me for a long time,
So I want to record my solution here (although it is now 2019, I don't know if anyone will still encounter this problem).
I followed the tutorial, changed the $CFG_QUE_TYPE from "Queue" to "Queue::POSIX" (my cpu is multi-core), and executed sphinxtrain run, which showed the same error as you.
Later, I changed the "Queue::POSIX" back to "Queue", and the problem disappeared.
I hope this answer will help the latecomers.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
1) Go to the database folder and run "python ../sphinxtrain/scripts/sphinxtrain -t an4 setup"
2) Look for etc/Sphinx_train.cfg and configure it. My sphinx_train.cfg configuration is in here.
3) Next step was to run "python ..\sphinxtrain\scripts\sphinxtrain run" but that does not work for me so I tried "python "D:/sphinx/sphinxtrain/scripts/sphinxtrain" run" and miraculously it run but an error was showing up. I attached below the screenshot of my failed training. I checked cmusphinx for this error and it says
"The .mfc files are the feature files converted from the input audio files on stage 000.comp_feats. Did you skip this step? Did you add new audio files without converting them? The training process expects a feature file to be there, and it isn't."
There was no step in http://cmusphinx.sourceforge.net/wiki/tutorialam that mentioned about 000.comp_feats. What step did I miss?
Last edit: Leimiaoren 2015-11-10
Your folder structure is wrong as I replied you in another thread. There is no need to start new thread for the same problem.
Stage comp_feats is described in tutorial, "Training internals" section
http://cmusphinx.sourceforge.net/wiki/tutorialam#training_internals
Thanks for patiently replying to my questions sir.
1) I followed what you said, I placed my database folder outside the sphinxtrain folder and now they are in one single folder. (For now I am using CMU's AN4 database, I will change this to my files later on when training the files are already succesful, I want to master the steps first)
2) I run "python ../sphinxtrain/scripts/sphinxtrain run"
3) It said configuration (sphinx_train.cfg) not defined but I am pretty sure I already configured it. But just to make it sure, I attached here my sphinx_train.cfg file.
4) I attached below the picture of the results when I run the line at #2.
Hoping to hear from you again since I am really illiterate when it comes to running the command prompt. Its my first time using it.
I am really sorry for creating multiple threads sir nickolay. I just need to train my database fast because I am running out of time.
According to the tutorial you need to run setup first with
Then you edit the configuration and run the training itself
I already did the setup. It said
"Sphinxtrain path: D:\sphinx\sphintrain
sphinxtrain binaries: D:\sphinx\sphinxtrain\bin\release
setting up the database AN4"
Then I configured the file and then run "python ../sphinxtrain/scripts/sphinxtrain run" ( I am working on windows)
then the error message occured as what I have attached above.
.
Last edit: Leimiaoren 2015-11-12
You can check contents of an4/etc folder to see if sphinx_train.cfg is there. Most likely it is not or it is corrupted by your edits.
I already followed all the steps you have given.
1) I redownload Sphinxtrain and AN4(So there would be no corrupt files), placed them inside one folder. Run the script "python ../sphinxtrain/scripts/sphinxtrain -t an4 setup"
2) Setting the database folder was successful, I then configured the sphinxtrain\etc\sphinx_train.cfg file.
3) Run "python ../sphinxtrain/scripts/sphinxtrain run" but still have the same problem. .MFC files are not being created by the scripts.
My logdir folder is empty so what is the problem?
Last edit: Leimiaoren 2015-11-13
This is wrong because you need to edit an4/etc/sphinx_train.cfg, not sphinxtrain/etc/sphinx_train.cfg
I suspect you also made many other mistakes. So it is hard to help you. I suggest you to find someone more experienced than you locally so he can help you.
Even if I replaced everything with new files. New sphinxtrain and new an4. Configured the one in an4\etc\sphinx_train.cgf (By the way, the CMU tutorial did not mention it was the file from an4, because there was also an etc file in the sphinxtrain folder so I just followed orders) the .MFC files are still not generating.
Sir can I ask how to install sphinxbase with the prebuilt binaries? or you don't have to install it anymore?
Hello Leimiaoren,
I had same problem and after this setup:
"python ../sphinxtrain/scripts/sphinxtrain -t an4 setup"
I only changed these lines and everything is ok:
$CFG_WAVFILES_DIR = "$CFG_BASE_DIR/wav";
$CFG_WAVFILE_EXTENSION = 'sph';
$CFG_WAVFILE_TYPE = 'nist'; # one of nist, mswav, raw
I have encountered the same problem as you.
This problem has been bothering me for a long time,
So I want to record my solution here (although it is now 2019, I don't know if anyone will still encounter this problem).
I followed the tutorial, changed the $CFG_QUE_TYPE from "Queue" to "Queue::POSIX" (my cpu is multi-core), and executed sphinxtrain run, which showed the same error as you.
Later, I changed the "Queue::POSIX" back to "Queue", and the problem disappeared.
I hope this answer will help the latecomers.