We want to inform you that we have been developing an Indonesian speech recognition using HTK and Julius for linux voice command and dictation. For now, only linux voice command is available. It is still in alpha stage, but quitely enough to operate computer hands freely.
It can be downloaded at http://groups.google.com/group/ilvc
But we are sorry, the documentation are still written in Indonesian. Feel free to ask me about the usage, or you can use google translator to translate into your own language.
ilvc only running on linux with gnome window manager. ilvc has some features such as:
*) operate gnome menu,
*) run dan close some favourites application such as open office, firefox, thunderbird, gedit, gimp, and totem and also it's menu operations
*) browse some pre-defined site, open a new tab, close tab, write words in textbox (i.e. google), etc.
*) mouse action (left-right, up-down, jump, diagonal moving, left-right click, etc)
*) linux addministration (pre-defined for Mandriva; or you can easily edit the normcmd.map to met your own linux distribution)
*) logout, reboot, shutdown
*) simple dictation (for digit to 99999, included the punctuations, space and another keyboard actions)
*) full dictation system (not available yet for public)
ilvc has a normalized command-map so it can easily ported to another language (ofcourse you must have your own accoustic model, grammar. dictionary and tiedlist). You just edit the langcmd.map in maps directory to use in your language.
Some example of langcmd.map dan normcmd.map are below:
## GNOME MENUS
buka menu aplikasi (English translate : open menu application) = SYSTEM_MENU_APPLICATIONS
buka menu (English translate : open menu) = SYSTEM_MENU_APPLICATIONS
jalankan firefox (English translate : open firefox) = OPEN_FIREFOX
jalankan internet (English translate : open internet) = OPEN_FIREFOX
batal (English translate : cancel) = CANCEL
tab = KEYBOARD_PRESS_TAB
mouse miring kanan atas (English translate : mouse diagonal up rightward) = MOUSE_DIAGONAL_UP_RIGHTWARD
mouse miring kanan bawah (English translate : mouse diagonal down rightward) = MOUSE_DIAGONAL_DOWN_RIGHTWARD
## GNOME MENUS
SYSTEM_MENU_APPLICATIONS = SYSTEM:MENU:OPEN:[S]APPLICATIONS
OPEN_FIREFOX = SYSTEM:APPCTL:OPEN:[S]/usr/bin/firefox
## KEYBOARD ACTIONS
CANCEL = SYSTEM:KEYBOARD:PRESS:[K]ESCAPE
KEYBOARD_PRESS_TAB = SYSTEM:KEYBOARD:PRESS:[K]TAB
# MOUSE ACTIONS
MOUSE_DIAGONAL_UP_RIGHTWARD = SYSTEM:MOUSE:MOVE:[M]250,-250,rel
MOUSE_DIAGONAL_DOWN_RIGHTWARD = SYSTEM:MOUSE:MOVE:[M]250,250,rel
You can edit or add the langcmd.map and normcmd.map to enlarge the ability of ilvc.
Terima kasih -Thank you very much
I just put a sample of ilvc for English voice command in http://groups.google.com/group/ilvc/files?upload=1 ilvc-en.tar.bz2
I used accoustic model and tiedlist from Voxforge Nigh Build HTK_AcousticModel-2008-09-21_16kHz_16bit_MFCC_O_D.tgz, made a sample ilvc-en.grammar and ilvc-en.voca.
Edit langcmd.map and oila... works like a charm.
This sample can run and close firefox, open office, gnome terminal and multimedia (totem). You can expand it by edit langcmd.map.
Here is a sample langcmd.map
OPEN MULTIMEDIA = OPEN_MUSICPLAYER
RUN MULTIMEDIA = OPEN_MUSICPLAYER
OPEN TERMINAL = OPEN_TERMINAL
RUN TERMINAL = OPEN_TERMINAL
OPEN ACROBAT = OPEN_ACROBAT
RUN ACROBAT = OPEN_ACROBAT
OPEN INTERNET = OPEN_FIREFOX
RUN INTERNET = OPEN_FIREFOX
OPEN FIRE FOX = OPEN_FIREFOX
RUN FIRE FOX = OPEN_FIREFOX
OPEN OFFICE = OPEN_OPENOFFICEWRITTER
RUN OFFICE = OPEN_OPENOFFICEWRITTER
CLOSE = SYSTEM_WINDOW_CLOSE
EXIT = SYSTEM_WINDOW_CLOSE