git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2810 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2809 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2808 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2807 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2806 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2805 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2804 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2794 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2791 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2789 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

trunk/src/nnet : bugfix in on-the-fly frame-shuffling, very long segments are now truncated correctly (hard limit on leftover to 1/2 of cachesize). Thanks to Nagendra for finding it!


git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2788 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

trunk:egs/wsj/s5/util merge subset_data_dir_tr_cv_spk.sh and subset_data_dir_tr90_cv10.sh to a single script subset_data_dir_tr_cv.sh

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2787 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

trunk:egs/wsj/utils/subset_data_dir_tr_cv_spk.sh (A data split script for cross validation based on random chosen speakers) -> modifications for better order of options and temporary files placement (Recommended by Karel Vesely)

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2786 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2784 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2783 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

trunk:egs/wsj/s5/utils - fix of typo for comment on subset_data_dir_tr_cv_spk.sh <= A data split script for cross validation based on random chosen speakers

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2782 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2781 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2779 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2776 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2774 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2773 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

improved configuration scheme


git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2772 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2771 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

trunk/egs/fisher_english: change fisher_data_prep.sh to support newer version of corpus (LDC, why must you constantly change directory names?)

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2770 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

This may be correct in some cases, when the *txt transcripts are not a part
of the BABEL delivery (as long as the STM file or RTTM file is)


git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2769 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2768 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

Use in cases when the *.txt files are not part of the delivery 
(such as evalpart1 subset delivery)


git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2767 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2766 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

This script can be called from any recognition script when needed.
This helps with future development and modification of the run* scripts


git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2765 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

command in the pipeline fails)


git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2764 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2763 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2762 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2761 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2760 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

- nnet-cache-tgtmat.cc : bugfix, 
- nnet-loss.cc : better code readability
- nnet-nnet.cc : refactored weights getter/setter, gradient getter
- nnet-rbm.h : for Gaussian-Bernoulli RBM training, enforce non-negative variance while comparing the variance of input-features/reconstruction-features, in anti-weight explosion detector (thanks to Daniel Korzinek)




git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2759 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2758 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2757 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2756 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2755 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2754 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8