Commit f6904fb9 authored by Dan Povey's avatar Dan Povey

Deleted stuff we don't need; minor fixes, some related to Valgrind issues and others cosmetic.

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@16 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8
parent e4616c96
......@@ -42,9 +42,8 @@ Other Source Material
current copyright holder, Henrique S. Malvar, has given his permission for the
release of this modified version under the Apache License 2.0.
This file includes material from the OpenFST Library v1.2.7 available at
http://www.openfst.org/twiki/bin/view/FST/WebHome and released under the
Apache License v. 2.0.
This project includes material from the OpenFST Library v1.2.7 available at
http://www.openfst.org and released under the Apache License v. 2.0.
[OpenFst COPYING file begins here]
......
This diff is collapsed.
These instructions are valid for UNIX-like systems (tested on
various Linux distributions; Darwin; Cygwin). For native Windows
These instructions are valid for UNIX-like systems (these steps have
been run on various Linux distributions; Darwin; Cygwin). For native Windows
compilation, see ../windows/INSTALL.
You must first have completed the installation steps in ../tools/INSTALL
(compiling OpenFst; getting ATLAS and CLAPACK headers).
You must first have completed the installation steps in ../tools/INSTALL or by
running ../tools/install.sh (compiling OpenFst; getting ATLAS and CLAPACK
headers).
The installation instructions are:
./configure
......@@ -14,6 +15,6 @@ make
Note that "make" takes a long time; you can speed it up by running make
in parallel, for instance "make -j 4"
For more information, see documentation at http://merlin.fit.vutbr.cz/kaldi/
For more information, see documentation at http://kaldi.sf.net/
and click on "The build process (how Kaldi is compiled)".
Example of debug command with valgrind:
in matrix/
valgrind ./matrix-lib-test
......@@ -13,15 +14,15 @@ looking at valgrind.out in the last directory, and looking for the last
filename *-test that was printed).
---
For emacs users:
ensure proper indentation it may be helpful to copy
http://google-styleguide.googlecode.com/svn/trunk/google-c-style.el
to your home directory as ".google-c-style.el", and add to your .emacs file the lines:
\verbatim
(load-file "~/.google-c-style.el")
(add-hook 'c-mode-common-hook 'google-set-c-style)
(add-hook 'c-mode-common-hook 'google-make-newline-indent)
\endverbatim
----
notes on potential install problems:
if you get errors like:
.../matrix/kaldi-matrix.cc:43: undefined reference to `clapack_sgetrf'
then you may be linking with a version of the library liblapack.a that
does not have this type of symbol: on UToronto's machines that library did
not export that symbol but there was a newer one called liblapack-3.a that
did.
If you get errors about not being able to find files in various system directories,
try typing 'make depend'.
--
......@@ -4,12 +4,14 @@
- Try to run the scripts and compile on various platforms, and
report problems.
TODO items (mainly for Dan to do, some for Arnab):
Check the clapack configuration in the configure script.
TODO items (Arnab):
Check the clapack configuration in the configure script... not clear
what the purpose of CLAPACK_ROOT is. I think the configure script
should create a Makefile that doesn't depend on external variables (Dan)
Add separate min-count at root of tree for regression-tree fMLLR/MLLR
Add fMLLR scripts for SGMM, and also use deeper trees.
Document configure script.
Add fMLLR scripts for SGMM.
TODO items (Dan):
Remove cpplint.py from distribution.
Remove feat/test_data/ (or Karel to get other data?).
Remove non-Kaldi code from decoder/
......
This source diff could not be displayed because it is too large. You can view the blob instead.
......@@ -24,8 +24,6 @@
- Subversion (svn): this is needed to download Kaldi and other software that it depends on.
- wget is required for the installation of some non-Kaldi components described below
- automake (and aclocal): required to install IRSTLM which some of our scripts use
(see below).
- doxygen: this is the program we use to generate html documentation from
the C++ source (which has been marked up according to doxygen requirements).
If you will be changing the source significantly, this might be useful.
......@@ -39,6 +37,8 @@
- IRSTLM: this a language modeling toolkit. Some of the example scripts require it but
it is not tightly integrated with Kaldi; we can convert any Arpa format
language model to an FST.
- The IRSTLM build process requires automake, aclocal, and libtoolize
(the corresponding packages are automake and libtool).
- sph2pipe: this is for converting sph format files into other formats such
as wav. It's needed for the example scripts that use LDC data.
- sclite: this is for scoring and is not necessary as we have our own, simple
......
......@@ -355,6 +355,7 @@ void ClusterGaussiansToUbm(const AmDiagGmm& am,
tmp_vars.CopyRowFromVec(tmp_vec, gauss_index);
gauss_index++;
}
DeletePointers(&(gauss_clusters_out[clust_index]));
}
tmp_gmm.Resize(opts.intermediate_numcomps, dim);
tmp_weights.Scale(1.0/tmp_weights.Sum());
......
......@@ -434,9 +434,8 @@ class CompartmentalizedBottomUpClusterer {
vector< vector<int32> > *assignments_out);
~CompartmentalizedBottomUpClusterer() {
for (vector< vector<Clusterable*> >::iterator itr = clusters_.begin(),
end = clusters_.end(); itr != end; ++itr) {
end = clusters_.end(); itr != end; ++itr)
DeletePointers(&(*itr));
}
}
private:
......@@ -568,7 +567,7 @@ bool CompartmentalizedBottomUpClusterer::CanMerge(int32 comp, int32 i, int32 j,
}
BaseFloat CompartmentalizedBottomUpClusterer::MergeClusters(int32 comp, int32 i,
int32 j) {
int32 j) {
KALDI_ASSERT(comp < ncompartments_ && i < npoints_[comp] && j < i);
clusters_[comp][i]->Add(*(clusters_[comp][j]));
delete clusters_[comp][j];
......
......@@ -113,6 +113,7 @@ BaseFloat ClusterBottomUp(const std::vector<Clusterable*> &points,
/// of compartments, such that only points in the same compartment are
/// clustered together. The compartment and pair of points with the smallest
/// merge cost are clustered.
/// The clusters in "clusters_out" are newly allocated and owned by the caller.
BaseFloat ClusterBottomUpCompartmentalized(
const std::vector< std::vector<Clusterable*> > &points, BaseFloat thresh,
int32 min_clust, std::vector< std::vector<Clusterable*> > *clusters_out,
......
......@@ -77,7 +77,7 @@ void TableEventMap::Write(std::ostream &os, bool binary) {
EventMap::Write(os, binary, table_[t]);
}
WriteMarker(os, binary, ")");
if(!binary) os << '\n';
if (!binary) os << '\n';
if (os.fail()) {
KALDI_ERR << "TableEventMap::Write(), could not write to stream.";
}
......@@ -174,8 +174,9 @@ std::string EventTypeToString(const EventType &evec) {
size_t EventMapVectorHash::operator ()(const EventType &vec) {
EventType::const_iterator iter = vec.begin(), end = vec.end();
size_t ans = 0; const size_t kPrime1=47087, kPrime2=1321;
for (; iter!= end; ++iter) {
size_t ans = 0;
const size_t kPrime1=47087, kPrime2=1321;
for (; iter != end; ++iter) {
#ifdef KALDI_PARANOID // Check names are distinct and increasing.
EventType::const_iterator iter2=iter; iter2++;
if (iter2 != end) { assert(iter->first < iter2->first); }
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment