<< go_back -Main Site- >> Configuration of Simon

Installation

To get speech recognition on your Linux box you need basicly four things:

A language/acoustic model that supports your spoken language
Simon, the speech recognition program
MarryTTS the spreech-synthesizer to answer you
A scenario for Simon, so some actions can be done

This document will guide you through the install process of may of the above. For the rest, see the Configuration page.
Everything (apart from Simon) is installed in .linjark in your HOME.

First we will download the linjark script, that in turn will download and install MarryTTS and set the groundwork for the modules later on.

Be aware that you still need to install Simon from your distribution repository (or compile it yourself).

Installation of linjark

For Marry to work, you need a function install of Oracle latest Java! (nothing else is tested)

Dependencies:

Linkajrk also depends on the following programs to function properly. Please install the necessary packages.
wmctrl, xdotool, html-xml-utils, kdialog, sox, xclip, curl
wmctrl xdotool html-xml-utils kdialog sox xclip curl

NOTE: Some modules maybe need additional software to function.

Download the Install.sh script and make it executable:
wget https://raw.githubusercontent.com/ManuGithubSteam/LinJarK/master/install.sh
chmod +x install.sh

Start the script with "i" parameter to start Install
./install.sh i

IMPORTANT:
Be aware that "/home/$USER/.linjark/bin" gets added to your $PATH variable.
Be sure that your .bashrc is called or .bash_profile gets executed on login, otherwise Simon will not function properly!

You can also add the Path to your /etc/environment. Check your distribution dokumentation for that!

You can also start Simon this way in a terminal:
PATH=\$PATH:/home/$USER/.linjark/bin simon

NOTE: If you are not happy with the predefined word who is used to address you (Sir or Miss) just edit .linjark/gender file and write in it what you want eg. Lord

Installation of Simon

Use the repository of your distribution to install Simon (min version: 0.4.1!).
It is a KDE/QT program and has KDE/QT dependencies among others used for speech model generation (sphinxtrain, sphinxbase etc).
Make sure that support for SHINX is present (default is HTK).

For Gentoo this would be:
USE Flags: kdepim, libsamplerate, nls, opencv, sphinx
emerge simon pocketsphinx sphinx3 SphinxTrain

For Debain/Ubuntu based distros:
apt-get install simon sphinx3 sphinxtrain

For Opensuse :
Add the repository: "OpenSUSE Build Service: KDE:Extras" (in Community Repositorys), then
zypper in simon
Also install SphinxTrain from software.opensuse.org:
http://software.opensuse.org/package/sphinxtrain?search_term=sphinxtrain

Also make sure that your microphone is properly configured and seen/heard by simon and alsa/pulseaudio etc. You can test this for example with Audacity :-D

For more information on Simon see:
https://userbase.kde.org/Simon
https://userbase.kde.org/Simon/Handbookk
https://forum.kde.org/viewforum.php?f=216

In general it is a very good idea to read the handbook! Simon must work correctly for the speech recognition.

If you compile it from source, make sure that +nls, +sphinx are compiled in!

Installation of the language model in Simon

This step will install the German language model who is provided by me.
If you want to use Simon with another language you can find language models for SHINX here. For further steps on this look into the FAQ.

Navigate in Simon to "Configure Acoustic Model" -> "Open Model" -> "Import"
Now navigate to you HOME there should be an voxforge*.sbm file. Import it.

Tick the box: "Adapt base model using training samples"

Next step: >> Configuration of Simon