Introduction to Kaldi Installation and compilation of virtual machine Ubuntu system

Kaldi Official website: http://www.kaldi-asr.org/doc/index.html

Download the Kaldi Zip package directly in Http://github.com/kaldi-asr/kaldi, without using git, the git download code is as follows:

1  git clone https://github.com/kaldi-asr/kaldi.git Kaldi--origin upstream

If the system does not have Git installed, install git first

1 sudo apt-get install git

Figure 1 Kaldi Code structure

The more important folders in Kaldi are EGS (example), SRC (source code), tools.

After downloading the Kaldi package, we also need to install some software tools under the system, I am using the virtual machine Ubuntu system, with the Apt-get tool, which can be used to download and install the software package from the authentication source.

1 sudoApt-getInstallSubversion2 sudoApt-getInstallAutomake3 sudoApt-getInstallLibtool4 sudoApt-getInstallzlib1g5 sudoApt-getInstallzlib1g-Dev6 sudoApt-getInstallRuby7 sudoApt-getInstalllibatlas-Dev8 sudoApt-getInstalllibatlas-base-Dev9 sudoApt-getInstall wget
Ten sudo apt-get install git

When the above packages are installed, additional dependencies are sometimes installed and installed automatically.

Kaldi Compile:

1. Go to Tools subdirectory, open terminal, run Make or Make-j 4 (multi-core parallel speed)

2. Go to src subdirectory, open terminal, run./configure

3. Make depend, check for dependent libraries

4. Run Make or Make-j 4 (multi-core parallel speed)

The above is the compiler execution process under the CPU. My computer is rotten, self-study tensorflow and Kaldi and other deep learning related are in the CPU test, first in-depth understanding of the use of tools and principles, later work in the GPU, run speed.

Introduction to Kaldi Installation and compilation of virtual machine Ubuntu system