If you were asked to design an Arabic speech recognition system for any application (e.g. automatic translation, video games) what variables do you need to bear in mind?
Try to work out technical and practical issues relating to your system and its potential users. In order to formulate relevant ideas and broaden your horizons, it is wise to familiarize yourselves with speech recognition systems available at your disposal such as your home computer, mobile device, vehicle telematics, etc.
There are many difficulties involved in designing Arabic speech recognition systems.
The first level of complexity for Arabic has to do with the many regional variants of the language. Would you design a system that would focus on Modern Standard Arabic? The Classical Arabic of the Koran? Or various different forms of colloquial Arabic? An ideal system would recognize the various different forms of Arabic, but that would be much harder to program than one that only recognized a specific dialect or asked the user to select from a menu of dialects.
You also would need to deal with general issues such as whether you want it to recognize all elements of the language or only certain commands and how to handle issues like background noise and variations among individual speakers.