Scope

The project will focus on developing the following systems.

1. Speech Recognition System for Urdu Language with the following specifications:

• A system that will recognize district names across Pakistan in six major Urdu accents (Sindhi, Punjabi, Seraiki, Balochi, Pashto and Urdu). The system will be limited to district names and will not recognize other location names, e.g. names of towns, villages, tehsils, etc. The system will recognize individual district names only and not complete sentential responses.
• A system that will recognize main location names of Lahore in Lahore sub-urban accent of Urdu. The recognition vocabulary will be limited to 200 major locations in Lahore. The system will recognize individual location names only and not complete sentential responses.

2. Text to Speech System for Urdu Language with the following specifications:

• A system that can read out Urdu text with a vocabulary of 10,000 words. The system will be able to process textual input and convert it into speech. It will implement basic declarative and interrogative intonation patterns.

3. Mobile Based Dialog System in Urdu:

• A system which integrates speech recognition system for district names to provide weather information for different districts in Pakistan
• A system which integrates speech recognition system for location names of Lahore city to provide location based services
• Generic dialog framework which can be extended for other services

4. Urdu Screen Reader:

• Localized open source screen reader. The screen reader will be able to read out Urdu text. It will not process English text as that can be done by using the existing screen readers. Urdu text to speech system will be wrapped into relevant Speech API (SAPI) for integration with the screen reader for this purpose.