Kaldi for Dummies Tutorial
Introduction
You will learn how to install Kaldi, how to make it work and how to run an ASR system using your own audio data.
environment
Rule number 1- use Linux. Although it is possible to use Kaldi on Windows.
Linux will do the job with the less amount of problems.
Some necessary stuff should be installed.
must have:
atlas
altas is a module for automation and optimization of calculations in the field of linear algebra,
autoconf
autoconf is a module for automatic software compilation on different operating systems,
automake
automake is a module for creating portable Makefile files,
git
git is a module for distributed revision control system,
libtool
libtool is a module fore creating static and dynamic libraries,
svn
svn is a module for revision control system (Subversion), necessary for Kaldi download and installation,
wget
wget is a module for data transfer using HTTP, HTTPS and FTP protocols,
zlib
zlib is a module for data compression,
probably has to be installed)
awk
awk is a programming language, used for searching and processing patterns in files and data streams,
bash
bash is Unix shell and script programming language,
grep
grep is command-line utility for searching plain-text data sets for lines matching a regular expression,
make
make is a module, which automatically builds executable programs and libraries from source code,
perl
perl is a dynamic programming language, perfect for text files processing.
download Kaldi
Follow the instruction carefully
http://kaldi-asr.org/doc/install.html
If you do not have much idea about how to use GIT, please read "Version control with Git".
http://kaldi-asr.org/doc/tutorial_git.html
you can install Kaldi in this directory (called 'Kaldi root path'):
/home/{user}/kaldi-trunk
Kaldi directories structures
Kaldi-trunk is the main Kaldi directory, and contains
egs: is example scripts to build ASR systems for over 30 speech corporas (documentation is attached for each project),
misc: is additional tools and supplies.
src: is Kaldi source code.
tools: contains some useful component and external tools.
windows: contains tools for running Kaldi using Windows.
Exemplary Project
Data preparation
Project finalization
Runing Scripts creation
Getting restuls
[0]