does not contain acoustic model definition 'mdef'

[sriram]

Hi,

I have trained the acoustic model with few sample sentences as per the instruction. The training is completed properly. But i couldnt run the decoder to test it. I am facing the following error.

ERROR: "acmod.c", line 83: Folder /model_parameters/7731.cd_cont_200' does not contain acoustic model definition 'mdef'.

I have attached the log file as well as the model configurations.

https://www.dropbox.com/s/84ubw18bstj73c7/TAR7731.zip?dl=0

Can anyone help me to sort out the issue.

Thanks and Regards,
Sriram.

[sriram]

When i have checked the log files of the previous steps, i got the following warning,

Phase 5: Determine amount of training data, see if n_tied_states seems reasonable.

Estimated Total Hours Training: 0.00346111111111111

This is a small amount of data, no comment at this time

Is this a cause for the issue?

[sriram]

The issue has been solved. I used Context independent model, so when the script searches by default for context dependent model in shinx_train.cfg, it failed to open the model directory. Then i changed the script to look for CI model, the issue got solved.

But still facing issue with high amount of error rate ;-)

[Sri Ningsih]

The issue has been solved. I used Context independent model, so when the script searches by default for context dependent model in shinx_train.cfg, it failed to open the model directory.

can you explain the detail, how can you used Context independent model? it must change in the sphinx_train.cfg?