Updates to branches/1, and minor changes to READMES + one #include statement.

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@138 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

misc/htk_conversion/README

@@ -1,4 +1,9 @@
 
+This is an example of a HTK conversion script that converts HTK models
+to Kaldi format models.  It's not being actively maintained.  This
+script is not very general-- since HTK's MMF format can contain almost
+anything, it's hard to create a general solution.
+
 # HTK conversion script:
 ./convert_htk.sh ../.. /mnt/matylda5/jhu09/setup/CH1/English/exp/xwrd.R0_800_TB500/hmm84/MMF /mnt/matylda5/jhu09/setup/CH1/English/exp/xwrd.R0_800_TB500/hmm10_800_500/cluster.trees convert_dir
 

misc/htk_decode_example/README

@@ -1,3 +1,6 @@
+# This script is not being maintained and is not guaranteed to work.
+# Its inputs are specific to the BUT system.
+
 
 mkdir -p feats
 dir=`pwd`

misc/htk_graph_creation_example/README

@@ -1,3 +1,7 @@
+This script is not being maintained and is not guaranteed to work.
+Its inputs are specific to the BUT system.
+
+
 
 This directory contains some scripts that convert HTK models into Kaldi format
 and build decoding graphs.  They take as input a CallHome English setup that's

src/TODO

@@ -17,7 +17,7 @@
 
  TODO items (Dan):
  [add ,s,cs on sgmm3e feats]
- remove back-compat options for gmm-est-et etc. (mean-and-var, etc.)
+
   Look into how better to estimate vectors in training time
     (I think WERs degraded when I fixed a script bug that
     was having the effect of re-estimating spk-vecs from

src/transform/hlda.h

@@ -23,7 +23,6 @@
 
 #include "base/kaldi-common.h"
 #include "matrix/matrix-lib.h"
-#include "optimization/optimization-common.h"
 #include "gmm/am-diag-gmm.h"
 #include "transform/transform-common.h"
 #include "transform/regression-tree.h"