Menu

sphinxtrain ERROR: "model_def_io.c" and senone

Help
ab1984
2015-08-07
2015-08-22
  • ab1984

    ab1984 - 2015-08-07

    I'm doing a speech recognition system for sinhala language. I have downloaded latest versions of sphinxbase,sphinxtrain and pocketsphinx and build on Ubuntu 14.04 virtual machine. Didn't get any errors during the installation. but when training the data I got this error repeatedly I have tried following,

    1. Reduce the phonemes to 55 (earlier it was 300 ++)
    2. Re record some of the errorness recordings.

    Even though their are no errors in the training iterations I got the following error at the end of training iterations,

    ERROR: "model_def_io.c", line 413: Unable to open /home/bopagama/Desktop/Workspace/sentence/model_architecture/sentence.200.mdef for reading: No such file or directory
    INFO: senone.c(210): Reading senone mixture weights: /home/bopagama/Desktop/Workspace/sentence/model_parameters/sentence.cd_semi_200_delinterp/mixture_weights
    FATAL: "senone.c", line 213: fopen(/home/bopagama/Desktop/Workspace/sentence/model_parameters/sentence.cd_semi_200_delinterp/mixture_weights,rb) failed

    I'm not using a big amount of training data. Is that the reason why I am getting this error?
    Or
    Did I do somthing wrong in setting up the files?

    Log Folder and my data files are attched Here
    Thank you

     
    • Nickolay V. Shmyrev

      For small amount of data for training without context-dependent models use continuous model instead of semi-continuous one and this error will be solved.

      This warning is caused by the fact you disabled context-dependent training and used semi-continuous model. You can ignore this error until you have more data and use context-dependent training.

       
      • ab1984

        ab1984 - 2015-08-19

        My final goal is to build voice recognition application for android mobile phones. In that case according to my understanding I should use semi-continuous mode. Am I correct?

        what is the difference between context dependent and context independent?

        thank you

         
  • ab1984

    ab1984 - 2015-08-07

    Thank You for your reply nickolay

    Now I got a different problem. Aligning the sound with my sentences.their are so many errors. So i decided to change the sphinx_train.cfg setting $CFG_FORCEDALIGN = 'no'; to yes acording to the tutorial at http://cmusphinx.sourceforge.net/wiki/tutorialam#troubleshooting and ran the train. Now i'm getting following error and I can't find any file named sphinx3_align in my system and also I used pocketsphinx, sphinxbase and sphinxtrain to build.

    MODULE: 11 Force-aligning transcripts
    Skipped: No sphinx3_align(.exe) found in /usr/local/libexec/sphinxtrain
    If you wish to do force-alignment, please copy or link the
    sphinx3_align binary from Sphinx 3 to /usr/local/libexec/sphinxtrain
    and either define $CFG_MODEL_DIR in sphinx_train.cfg or
    run context-independent training first.

    Did I do any wrong when building sphinxtrain?
    I just follow the tutorial and issue following commands

    ./autogen.sh
    make
    make install

    Thank You

     
  • ab1984

    ab1984 - 2015-08-07

    I got it corrected I think !

    I seperately install sphinx3 and copied the sphinx3_align file to sphinxtrain folder and ran the sphinx train.
    Am I doing it correctly?

    I have attached the log directory for your reference. I'm still getting following error,

    **ERROR: FATAL: "main.c", line 730: Initialization failed
    Skipping SIL
    MODULE: 45 Prune Trees
    Phase 1: Tree Pruning
    ERROR: FATAL: "main.c", line 167: Unable to open /home/bopagama/Desktop/Workspace/sentence/trees/sentence.unpruned/AI-0.dtree for reading: No such file or directory
    MODULE: 50 Training Context dependent models
    Phase 1: Cleaning up directories:
    accumulator...logs...qmanager...
    Phase 2: Copy CI to CD initialize
    ERROR: This step had 1 ERROR messages and 0 WARNING messages. Please check the log file for details.
    Phase 3: Forward-Backward
    Baum welch starting for 1 Gaussian(s), iteration: 1 (1 of 1)
    0% ERROR: FATAL: "main.c", line 1846: initialization failed

    ERROR: This step had 1 ERROR messages and 0 WARNING messages. Please check the log file for details.
    ERROR: Failed to start bw
    ERROR: Only 0 parts of 1 of Baum Welch were successfully completed
    ERROR: Parts 1 failed to run!
    Training failed in iteration **

    Any idea how should i proceeed?

    Thank You

     
    • Nickolay V. Shmyrev

      This particular phone 'AI' is never observed in your database, so you need to remove it from the dictionary and phoneset. You can replace it with more common phone.

       
    • erkan sahin

      erkan sahin - 2015-08-22

      did you solve problem?

       

Log in to post a comment.