Dear Mohamed Magdy,
>> This sounds great! How usable is it?
I did not test it for online, but it works fine with offline text, but i need to confirm the corpus was limited as well as the training data, and that was stop me from continuing the Master as I could not get enough spoken sentence to train and test with.
Also I forget to mention my corpus is Egyptian Colloquial Arabic.
But the good news is if we have a corpus or a sentence for any other language it will be systematic steps to build a model or adapt the model to the Colloquial Language.
>> 1- I have some interested at this field, but did you try to upload your material to Arabisc
or >> some other open source.
>> I think this is the suitable place for such materials?
Ok no problem do you know who is Arabisc owner to discuss this with him, i checked arabisc it seems interesting but i have different apporach
Arabisc is an ASR Engine but I need to build a system to ease the building of ASR System + the ASR Engine itself, by just adding some text files and some spoken sentence as wave file then run it, then you have it, so I need to upload not just the monophones and triphones model i got I need to upload the scripts and the tools i used to build it and the tools to test it
2- Also you can write some brief for the materials at wiki.arabeyes.org
Ok I will do it as soon as possible In Shaa; Allah
3- Is your OCR project will be open source?
In Shaa' allah, but it will depend on if i will get some modules from other research e.g. the feature extractor which i currently stuck in it, if I borrow the feature extractor from other research I can not put it as open source until I got an approval from the owner.
4- could you please add some info about you here
بيانات المهتمين بحوسبة اللغة العربية
Thanks I will check it out.