I successfully trained a UBM model and the user specific GMMs using that UBM model.
But while prediction, when I call predict_one_with_rejection() in gmmset.py, It shows me SEGMENTATION FAULT.
NOTE:
Segmentation fault occurs while generating scores from UBM Model. I have gone through the code myself and found that scores can be generated successfully when GMM models are used to call the function gmm_score(self, gmm, X) in gmmset.py file. But as soon as the object of an UBM model is passed, it throws SEGMENTATION FAULT with no other error.
The segmentation fault comes from the shared library, 'pygmm.so' so I can't track it down further. But from what I found is, the function called from that point leads to a function score_batch(GMM *gmm, double *X_in, double prob_out, int nr_instance, int nr_dim, int concurrency) in the file pygmm.cc in gmm/src/.
I successfully trained a UBM model and the user specific GMMs using that UBM model. But while prediction, when I call predict_one_with_rejection() in gmmset.py, It shows me SEGMENTATION FAULT.
NOTE: Segmentation fault occurs while generating scores from UBM Model. I have gone through the code myself and found that scores can be generated successfully when GMM models are used to call the function gmm_score(self, gmm, X) in gmmset.py file. But as soon as the object of an UBM model is passed, it throws SEGMENTATION FAULT with no other error. The segmentation fault comes from the shared library, 'pygmm.so' so I can't track it down further. But from what I found is, the function called from that point leads to a function score_batch(GMM *gmm, double *X_in, double prob_out, int nr_instance, int nr_dim, int concurrency) in the file pygmm.cc in gmm/src/.
Please explain the reason for the same.