oazabir / Quran

Quran Research Tool
77 stars 46 forks source link

Phonetic search capability #4

Closed foyzulkarim closed 12 years ago

foyzulkarim commented 12 years ago

Create a phonetic index of the Arabic and the English and expose that through the search. If someone searches using qalu, it will match qaloo, yaqooloo etc forms of the same word. It will take the English form and automatically detect the appropriate arabic form and search using that arabic form.

oazabir commented 12 years ago

I am curious to know your algorithm. Let me know how you plan to create the phonetic index.

foyzulkarim commented 12 years ago

I had an opportunity to do some R&D on a CMU Sphinx and Arabic Automatic Speech Recognition (AASR). My plan is to get help using these two libraries and come up with an algorithm. I also tested Microsoft's speech recognition library but it is not good enough and we can't change that library according to our need.

oazabir commented 12 years ago

Excellent. some time back I saw this: http://www.codeproject.com/Articles/26880/Arabic-Soundex

May be this will help.

foyzulkarim commented 12 years ago

Should the software be able to detect the words from voice?

oazabir commented 12 years ago

No, that will be a never ending research project.

Soundex isn't about voice recognition. It converts arabic words into a representation that is same for different spelling of the same word. So, qalu, qaloo, qaaloo all will result in say q0l1. This way you can take user's incorrect spelling and find similar matches.

oazabir commented 12 years ago

https://github.com/oazabir/Quran/wiki/Database-Design

foyzulkarim commented 12 years ago

I have perform some practices on the Soundex and Arabic-Soundex codes. Is there any existing GUI or mockup for this feature? Or should I propose a detailed feature of this?

oazabir commented 12 years ago

There's no page now. Please add a new say Search.aspx and implement your code there. We will later on merge it with the homepage.

oazabir commented 12 years ago

Hi, how are you doing with the search?

foyzulkarim commented 12 years ago

I paused working for last couple of days. Sorry. Will resume by today.

oazabir commented 12 years ago

This issue is now abandoned as the work will not be continued.