Help on Arabic Speech Recognition

[Muscha]

Hi. everybody

i am newbie in sphinx.i want to create Arabic speech recognition application
(voice to text and text to voice)

i have download all the latest sphinx4 package,sphinx train,sphinx3 and
reading all the documentation in cmu website
but i still confused n stuck..
i have question and i need ur suggestion guys

1. how to create dictionary and language model in Arabic language?is it can be done it this site? --->http://www.speech.cs.cmu.edu/tools/lmtool.html
2. what are the steps that should i do to create Arabic speech recognition?either i need to record the voice first or create dictionary
3. how to do acoustic model using Arabic language?is it same with English language?
   can u guys give me the tutorial or link that can be my resource..

Thx for help, i really appreciated

regard

[Nickolay V. Shmyrev]

1. how to create dictionary and language model in Arabic language?is it can
   be done it this site? http://www.speech.cs.cmu.edu/tools/lmtool.html

No, you need to do it by hand. You can write initial dictionary yourself by
hand. Read for example

http://www.dur.ac.uk/daniel.newman/phon5.pdf

To create language model from text you can use cmuclmtk

1. what are the steps that should i do to create Arabic speech
   recognition?either i need to record the voice first or create dictionary

The same as described in tutorial

http://cmusphinx.sourceforge.net/wiki/tutorialam

1. how to do acoustic model using Arabic language?is it same with English
   language?

yes

[Nickolay V. Shmyrev]

And, actually there is

https://sourceforge.net/projects/arabisc/

Which you can just download

[Muscha]

Tq for replying Nickolay
i'll try to do what u suggested,but if i hv mistake i will ask again :D

To create language model from text you can use cmuclmtk

i have read this site --->http://cmusphinx.sourceforge.net/wiki/cmuclmtkdevel
opment
it is seem that i need to do manually

i already install cygwin but i can use the configuration based on this site
--> http://cmusphinx.sourceforge.net/wiki/cmuclmtkdevelopment

is there any way that i can create Arabic language model without using
cmuclmtk?
or is there any site provide it?

Tq

Regard

[hiyassat]

do you use arabic text ?
send your text to me nd i will make LM for you
also if you can upload your wave or features file i can help you in developing
acoustic model

[Muscha]

do you use arabic text ? send your text to me nd i will make LM for you also
if you can upload your wave or features file i can help you in developing
acoustic model

yes i do
my application will input voice and convert it to text.. & input text will
recognize the Arabic voice (text to voice)

these are my text

الى اين انت ذاهب؟
انا ذاهب الى المدرسة
كتاب
قلم
مدرسة
كمبيوتر
موسيقى
اسود
ابيض
احمر
اصفر

What do you mean feature file?it is config.xml?

Thank you for reply, i will upload voice as soon as i can..

[B_Haddad]

Can you give us a tutor on how to use this tool

[B_Haddad]

I mean the Dialog tool

[hidayu kamarudin]

yes i really do hope too can someone teach on how to use the dialog tool?

[maisara yasser]

salam all
i work in Arabic speech recognition , i have a dictionary but i don't know how built language model and acoustic model can anyone tell me the steps for building the system
thanks

[maisara yasser]

how can use arabisc for acoustic model?

[oumlghayth]

hello everybody,
I'm trying to use sphinx to segment my arabic corpus using forced alignement, and i'm new to this toolkit , please how can I use it ?
what are the steps ?
thanks for your help :)

[oumlghayth]

Hello,
I would like to mention that i have 3.5 hours of speech with the orthographic transcription
And I want to segment it by placing the boundries automatically in the center of the vowel.
I don't know how to prepare files for training...
Thx for your help

Regards