Menu

new to sphinx4 - problems while training

Help
tho_mee
2009-01-01
2012-09-22
  • tho_mee

    tho_mee - 2009-01-01

    I am totally new to sphinx...Hope someone can help me out.

    I downloaded SphinxTrain trainer and an4_sphere database. I have copied both to a folder and have run the following.
    cd SphinxTrain
    ./configure
    make

    perl scripts_pl/setup_tutorial.pl an4

    cd an4
    perl scripts_pl/make_feats.pl -ctl etc/an4_train.fileids
    till now there is no problem....

    After this rather than running all scripts together i am running one by one and got the following error messages in between when doing the training at module 30.
    I got this when gave

    perl scripts_pl/30.cd_hmmuntied/slave_convg.pl

    Following was the details given in an4.html
    What is the problem? what should i do?
    *********
    MODULE: 30 Training Context Dependent models (2009-01-01 13:20)

    Phase 1: Cleaning up directories:
    accumulator... logs... qmanager... completed

    Phase 2: Initialization

    mk_mdef_gen Log File

    WARNING: This step had 0 ERROR messages and 1 WARNING messages. Please check the log file for details.
    completed

    init_mixw Log File
    completed

    Phase 3: Forward-Backward

    Baum welch starting for iteration: 1 (1 of 1)

    bw Log File
    completed

    Normalization for iteration: 1

    norm Log File
    completed

    Current Overall Likelihood Per Frame = 0.353931891107765

    Baum welch starting for iteration: 2 (1 of 1)

    bw Log File

    <b>This step had 2 ERROR messages and 1 WARNING messages. </b>Please check the log file for details.
    completed

    Normalization for iteration: 2

    norm Log File

    WARNING: This step had 0 ERROR messages and 3 WARNING messages. Please check the log file for details.
    completed

    Current Overall Likelihood Per Frame = 3.2847686038117

    Convergence Ratio = 8.28079296141062

    Baum welch starting for iteration: 3 (1 of 1)

    bw Log File

    <b>This step had 2 ERROR messages and 1 WARNING messages. </b>Please check the log file for details.
    completed

    Normalization for iteration: 3

    norm Log File

    WARNING: This step had 0 ERROR messages and 3 WARNING messages. Please check the log file for details.
    completed

    Current Overall Likelihood Per Frame = 4.21795544474383

    Convergence Ratio = 0.284095153567056

    Baum welch starting for iteration: 4 (1 of 1)

    bw Log File

    This step had 2 ERROR messages and 1 WARNING messages. Please check the log file for details.
    completed

    Normalization for iteration: 4

    norm Log File

    WARNING: This step had 0 ERROR messages and 3 WARNING messages. Please check the log file for details.
    completed

    Current Overall Likelihood Per Frame = 4.34042914120474

    Training completed after 4 iterations

     
    • tho_mee

      tho_mee - 2009-01-06

      hi..

      i recorded my voice as .wav files and made the above mentioned changes in config file . But some error is coming during feature extraction
      i am posting the same below.

      -cfg not specified, using the default ./etc/sphinx_train.cfg
      -param not specified, using the default ./etc/feat.params
      bin/wave2feat \
      -verbose yes \
      -alpha 0.97 \
      -dither yes \
      -doublebw no \
      -nfilt 40 \
      -ncep 13 \
      -lowerf 133.33334 \
      -upperf 6855.4976 \
      -nfft 512 \
      -wlen 0.0256 \
      -c etc/ncst_train.fileids \
      -mswav yes \
      -di /home/ahmad/sphinx/ncst/wav \
      -ei wav \
      -do /home/ahmad/sphinx/ncst/feat \
      -eo mfc

      [Switch] [Default] [Value]
      -help no no
      -example no no
      -i
      -o
      -c etc/ncst_train.fileids
      -nskip
      -runlen
      -di /home/ahmad/sphinx/ncst/wav
      -ei wav
      -do /home/ahmad/sphinx/ncst/feat
      -eo mfc
      -nist no no
      -raw no no
      -mswav no yes
      -input_endian little little
      -nchans 1 1
      -whichchan 1 1
      -logspec no no
      -feat sphinx sphinx
      -mach_endian little little
      -alpha 0.97 9.700000e-01
      -srate 16000.0 1.600000e+04
      -frate 100 100
      -wlen 0.025625 2.560000e-02
      -nfft 512 512
      -nfilt 40 40
      -lowerf 133.33334 1.333333e+02
      -upperf 6855.4976 6.855498e+03
      -ncep 13 13
      -doublebw no no
      -warp_type inverse_linear inverse_linear
      -warp_params
      -blocksize 200000 200000
      -dither yes yes
      -seed -1 -1
      -verbose no yes
      INFO: fe_interface.c(100): You are using the internal mechanism to generate the seed.
      INFO: fe_sigproc.c(752): Current FE Parameters:
      INFO: fe_sigproc.c(753): Sampling Rate: 16000.000000
      INFO: fe_sigproc.c(754): Frame Size: 410
      INFO: fe_sigproc.c(755): Frame Shift: 160
      INFO: fe_sigproc.c(756): FFT Size: 512
      INFO: fe_sigproc.c(757): Lower Frequency: 133.333
      INFO: fe_sigproc.c(758): Upper Frequency: 6855.5
      INFO: fe_sigproc.c(759): Number of filters: 40
      INFO: fe_sigproc.c(760): Number of Overflow Samps: 0
      INFO: fe_sigproc.c(761): Start Utt Status: 0
      INFO: fe_sigproc.c(763): Will add dither to audio
      INFO: fe_sigproc.c(764): Dither seeded with -1
      INFO: fe_sigproc.c(771): Will not use double bandwidth in mel filter
      INFO: wave2feat.c(139): /home/ahmad/sphinx/ncst/wav/are.wav
      LENGTH: 4
      INFO: wave2feat.c(786): Reading MS Wav file /home/ahmad/sphinx/ncst/wav/are.wav:
      INFO: wave2feat.c(787): 16 bit PCM data, 2 channels 84708 samples
      INFO: wave2feat.c(788): Sampled at 16000
      ERROR: "wave2feat.c", line 883: unknown input file format
      ERROR: "wave2feat.c", line 201: error reading speech data
      FATAL_ERROR: "wave2feat.c", line 90: error converting files...exiting

      Where did I go wrong??

       
    • Nickolay V. Shmyrev

      > What is the problem? what should i do?

      Continue with training? There is no problem in the messages you provided.

       
    • tho_mee

      tho_mee - 2009-01-01

      k.
      i continued and at MODULE: 50 Training Context dependent models also i got some errors. There is no problem with that also??? I have copied a part of the error message below...

      One more thing i wanted to know...
      Is it possible to place .wav files recorded by me in the an4/wav folder( which now has .sph files) and do this training and get a speech recognition system which recognizes my sound?


      Current Overall Likelihood Per Frame = 2.67914672700468

      Convergence Ratio = 0.0314306119339485

      Baum welch starting for 2 Gaussian(s), iteration: 1 (1 of 1)

      bw Log File

      WARNING: This step had 0 ERROR messages and 1 WARNING messages. Please check the log file for details.
      completed

      Normalization for iteration: 1

      norm Log File

      This step had 34 ERROR messages and 0 WARNING messages. Please check the log file for details.
      completed

      Current Overall Likelihood Per Frame = 2.19669113299328

      Baum welch starting for 2 Gaussian(s), iteration: 2 (1 of 1)

      bw Log File
      completed

      Normalization for iteration: 2

       
      • Nickolay V. Shmyrev

        > This step had 34 ERROR messages and 0 WARNING messages. Please check the log file for details. completed

        These errors are errors in the data processing and not in the training. It was discussed so many times on this forum you can just search. They means your data is not consistent with transcription you provided.

        > Is it possible to place .wav files recorded by me in the an4/wav folder( which now has .sph files) and do this training and get a speech recognition system which recognizes my sound?

        Yes, you just need to change file format in sphinx_train.cfg to mswav instead of sph:

        $CFG_WAVFILE_EXTENSION = 'wav';
        $CFG_WAVFILE_TYPE = 'mswav'; # one of nist, mswav, raw

         

Log in to post a comment.

Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.