VoxForge
Hello, i'm new in speech recognition tecnology..
I'm italian student and work for University of Unisannio.
I'm working on the realization of a speech recognition tool, to facilitate the attendance of courses to students hearing.
The core of my project is Julius 4.1.3 ..
tell me someone would know how to use jcontrol, adintool and Julius in server mode?
If i run Julius with this line (windows):
julius-4.1.3 -input mic -C config.jconf -module
on server, and jclient with this line (windows):
jcontrol.exe ip_of_server
each process run, but.. The recognition is detected on server's microphone!
I serve the development from the client's microphone.
If i run Julius with this line:
julius-4.1.3 -C config.jconf -module
And jcontrol with line:
jcontrol.exe ip_of_server
Each process run and the connection is estabilished, but no computation is done (no recognition).
I have see more material on internet and I understand I have to use adintool to manipulate stream of data between jclient and Julius..
Can anyone suggest a tutorial, or show me some example code?
Thanks in advance for the time you have dedicated to me!
C.
SOLVED.
If you wont to use client for voice input-recognition text, and server to elaborate speech recognition, you have to run Julius, jcontrol and adintool so:
1- Start Julius (on server computer):
julius-4.1.3 -C conf.jconf -input adinnet -module
for completeness, conf.jconf contain:
## Grammar definition file (DFA and dictionary)
-dfa demo.dfa
-v demo.dict
## Acoustic HMM file
-h hmm15/hmmdefs
## HMMList that maps logical triphone to physical ones.
-hlist tiedlist
-smpFreq 48000