Click here to register.

General Discussion

Language recognition
User: f4bxw
Date: 11/19/2007 11:38 am
Views: 4324
Rating: 20


congratulations for this valuable project!

I'd like to start experimenting with voice processing in order to build an automated language recognition engine.

I'm very new to this area but I can imagine a simple processing system that performs phoneme extraction from voice file, and then use an n-gram regognition system. Is the "accoustic model" of a language the right starting point for my approach?



--- (Edited on 11/19/2007 11:38 am [GMT-0600] by f4bxw) ---

Re: Language recognition
User: nsh
Date: 11/19/2007 12:58 pm
Views: 267
Rating: 18

I suggest you to get some background in ASR first, read Spoken Language Processing by Huang and others:

 Rabiner's tutorial or even the HTK book.


--- (Edited on 11/19/2007 12:58 pm [GMT-0600] by nsh) ---

Re: Language recognition
User: kmaclean
Date: 11/19/2007 1:43 pm
Views: 256
Rating: 26

Hi Vincent, 

Another good reference/textbook is: 

SPEECH and LANGUAGE PROCESSING: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition, By  Daniel Jurafsky and  James H. Martin,

The second edition draft chapters are available for download in pdf format.

You might also look through the archives of comp.speech.research and/or the HTK email archives (you need to be registered with HTK) to see if anyone has already done some work in this area.

Good luck! 



--- (Edited on 11/19/2007 2:43 pm [GMT-0500] by kmaclean) ---

Re: Language recognition
User: DavidGelbart
Date: 11/20/2007 1:21 pm
Views: 749
Rating: 32

There has been quite a bit of research on this.  I suggest you do Google and Google Scholar searches for "language identification" and "language recognition".

Hope this helps,

--- (Edited on 11/20/2007 1:21 pm [GMT-0600] by DavidGelbart) ---

Re: Language recognition
User: dara
Date: 1/18/2010 8:19 am
Views: 124
Rating: 2


I am doing a research on language identification for speech. I have read some papers talking about several approaches I can do. One of them is using GMM (Gaussian Mixture Model). Nevertheless, I have no idea how to apply it to a real experiment. I want to run a small experiment relating to language identification by using GMM. Is there any tutorial I can follow, just like that of speech recognition tutorial on voxforge? I need to experience how to do the experiment. Thanks a million.




--- (Edited on 1/18/2010 8:19 am [GMT-0600] by Visitor) ---

Re: Language recognition
User: nsh
Date: 1/19/2010 4:01 am
Views: 1141
Rating: 2

For language recognition as well as speaker identificaitona or gender one, generic GMM toolkit could be used. Mistral

is a good one

--- (Edited on 1/19/2010 13:01 [GMT+0300] by nsh) ---