kaldi/tools/INSTALL

To install the prerequisites for Kaldi, try running
 ./install.sh
It tries to install various things, and it will tell you at the
end whether each item succeeded or failed.
If you get warnings that certain things failed to install but
it says "optional anyway", this does not matter.  But if it
just says "Failure" for one of the prerequisites, that is a problem--
you will have to go to the manual instructions in this file (below),
and try to install the prerequisite yourself and if you can, fix the
problem.

[note: ./install_atlas.sh is a backup plan in case you don't have ATLAS
 installed on your system; ignore it unless you get errors from
 ../src/configure.]

---

See below install instructions for:
(1) sph2pipe (needed to run the example scripts but not for compilation)
(2) openfst (needed)
(3) dot (optional; only used by doxygen and for FST debugging).
(4) IRSTLM (optional; only needed if you want to build LMs and
  don't already have a setup).
(5) sclite (optional; useful for detailed scoring output but the
   default scripts don't use it).
(6) ATLAS (needed only for headers, if already on system, but for
  native Windows compilation without Intel MKL you have to compile this)
(7) CLAPACK headers (required if you have the library available but
 no headers in a directory accessed by default; this is the case on Cygwin)
(8) libportaudio (needed for the online recognition binaries)


####

(1)
Install instructions for sph2pipe_v2.5.tar.gz

wget ftp://ftp.ldc.upenn.edu/pub/ldc/misc_sw/sph2pipe_v2.5.tar.gz
tar -xovzf sph2pipe_v2.5.tar.gz 
cd sph2pipe_v2.5
gcc -o sph2pipe *.c -lm 
cd ..

# These instructions are not valid for native Windows;
# see the readme, 0readme.1st, in sph2pipe_v2.5

####
(2)
Install instructions for OpenFst

Note that this should be compiled with g++-4.x
You may have to install this and give the option CXX=<g++-4-binary-name>
to configure, if it's not already the default (g++ -v will tell you).
(on cygwin you may have to install the g++-4.0 package and give the options CXX=g++-4.exe CC=gcc-4.exe to configure).


wget http://openfst.cs.nyu.edu/twiki/pub/FST/FstDownload/openfst-1.2.10.tar.gz
tar -xovzf openfst-1.2.10.tar.gz   
for dir in openfst-1.2.10/{src/,}include/fst; do
    ( [ -d $dir ] && cd $dir && patch -p0 -N <../../../../openfst.patch ) 
done 
rm openfst 2>/dev/null # Remove any existing link
ln -s openfst-1.2.10 openfst
 
cd openfst-1.2.10
# Choose the correct configure statement:

Linux or Darwin:
  ./configure --prefix=`pwd` --enable-static --disable-shared 

Linux, cross-compile (64-bit) [you probably don't want this]:
 # Configure for linux, cross-compile to 64-bit
 # Note that this refers to 64-bit compilation, which
 # on the BUT machines it not the default even though the machines
 # are 64bit [you would have to compile with OPT=64bit].  
  ./configure --host=x86_64-linux --prefix=`pwd` --enable-static --disable-shared 

Cygwin:
  ./configure --prefix=`pwd` CXX=g++-4.exe CC=gcc-4.exe --enable-static --disable-shared  

# make install is equivalent to "make; make install"
make install

####

(3) dot

This is used by doxygen to automatically generate documentation for the source, and
is also useful for debugging FSTs (in conjunction with fstdraw).

We provide here instructions for installing dot on cygwin.
On most native UNIX systems, dot is likely to be installed already.
If not, once you work out how to install it please put the instructions here.

Get the Windows version of dot from
http://www.graphviz.org/pub/graphviz/stable/windows/graphviz-2.26.3.msi
follow the default installation setup.
in your .bashrc file, add the line
export PATH=$PATH:/cygdrive/c/Program\ Files/Graphviz2.26.3/bin/
[you might want to first make sure that that's where you installed it.
If your shell is not bash, this may have to be changed a bit].

####

(4)
The following are instructions to install IRSTLM.
This is not needed for the basic system builds (RM has its own
non-ARPA LM, and WSJ comes with LMs).  So installing this may be
left till later, if you are in a hurry.

svn co https://irstlm.svn.sourceforge.net/svnroot/irstlm/trunk irstlm
cd irstlm
[
if your aclocal and automake are too old, you may want to 
adjust the settings for the variables ACLOCAL and AUTOMAKE
in ./regenerate-makefiles.sh.  For example, on cygwin you have
to install the package automake-1.9, and the two lines in
regenerate-makefiles.sh become:
ACLOCAL=/bin/aclocal-1.9
AUTOMAKE=/bin/automake-1.9 
... however, the defaults may work.
]
./regenerate-makefiles.sh  
./regenerate-makefiles.sh  # run this twice; it seems to be necessary.
./configure --prefix=`pwd`
#or for 64 bit cross-compilation (tested on BUT machines)
#./configure --prefix=`pwd` CC="x86_64-linux-gcc" CXX="x86_64-linux-g++" AR="x86_64-linux-ar" AS="x86_64-linux-as" RANLIB="x86_64-linux-ranlib"
[ you may have to install zlib before typing make ]
make
make install

####

(5) sclite [OPTIONAL!]  This can be helpful helpful for scoring but the default
scoring scripts do not use it (they use our own Kaldi-based scorer). 

If you get a bug about strncasecmp, you have to modify the file 'sctk-4.0/src/rfilter1/makefile'.
There is a comment in there about setting OPTIONS to be empty.  Do that and it
should compile. [Do this after "make config" but before "make install"].

Install instructions for sclite/sctk.
This was obtained as follows:

 (a) download it.
 If the following command fails with e.g. a proxy error:
   wget ftp://jaguar.ncsl.nist.gov/pub/sctk-2.4.0-20091110-0958.tar.bz2
 then try this:  
  ftp jaguar.ncsl.nist.gov 
  use username="anonymous" and an empty password
  cd pub
  get sctk-2.4.0-20091110-0958.tar.bz2
(b)
# Convert into gzip format (bunzip2 needs to be installed for this to work):
bunzip2 sctk-2.4.0-20091110-0958.tar.bz2
gzip sctk-2.4.0-20091110-0958.tar

tar -xovzf sctk-2.4.0-20091110-0958.tar.gz 
cd sctk-2.4.0
for x in src/asclite/core/recording.{h,cpp}; do # Fix a compilation error that can occur with newer compiler versions.
  sed 's/Filter::Filter/::Filter/' $x > tmpf; mv tmpf $x;
done
make config
make all
make check
make install
make doc

####

(6) Atlas
wget http://sourceforge.net/projects/math-atlas/files/Stable/3.8.3/atlas3.8.3.tar.gz
tar -xovzf atlas3.8.3.tar.gz ATLAS/include
# don't do any more installation at this point, we just need this for headers
# (except for native Windows compilation, for which see ../windows/INSTALL.atlas)

#####

(7) CLAPACK headers (in case needed, if CLAPACK is installed on the machine...
   we don't go to the trouble of installing CLAPACK if the libraries are not
   already there).
mkdir CLAPACK_include
cd CLAPACK_include
for x in  clapack.h	f2c.h cblas.h; do 
  wget http://www.netlib.org/clapack/$x; 
done

####

(8) Install instructions for libportaudio.

libportaudio is only needed for the online recognition binaries. It enables audio
capture from the sound card. Unfortunately, the installer for libportaudio may not 
work on all version of Linux or Mac OS. However, for most people it will be 
sufficient to simply run the script

./install_portaudio.sh

We tested this installation script on various versions of Suse Linux and Red Hat as
well as on Mac OS (Darwin). Please note that the installation script patches up the 
default Makefile of portaudio when installing on Mac (for details, please have a look
at the installation script itself). 

!!IMPORTANT!! UNDER MAC OS, YOU MAY NEED TO COPY THE libportaudio.dylib TO
/usr/local/lib SO THAT THE BINARIES CAN FIND THE LIBRARY. AFTER COMPILING 
LIPPORTAUDIO, YOU CAN FIND THE DYLIB FILE(S) IN portaudio/install/lib

The best bet for support on compiling libportaudio, should you face any problems, 
would be the portaudio documentation:

http://portaudio.com/docs/v19-doxydocs/tutorial_start.html

We know of at least one instance were libportaudio  compiled fine, but only 
produced garbage audio samples. This happened on a Linux system  that only had 
Open Sound System (OSS) installed. We therefore recommend to install ALSA. This 
is also reflected in our Makefile for the binaries in /src/onlinebin where we 
include -lasound.