Ongoing

Rehnuma Awaz: AI-Powered Urdu Screen Reader for Blind Users in Pakistan

Services Covered:

AI and Machine Learning • AI-Driven Data Solutions • Software Maintenance and Support • iOS Application Development • Android Application Development

Industry

Information Technology

33 months

Budget

1M - 10M

Start Date

1 January 2024

Team Size

16-20

Engagement Model

Onshore

Project Capability Score (PCS)

N/A

Not Available

Project Capability Score (PCS) is not available for this project because it has not been endorsed yet. Once an endorsement is added, the score can be calculated based on project feedback, company track record, experience, and recency. Learn more about PCS

Client

IGNITE National Technology Fund

Islamabad, Pakistan

Project Summary

Rehnuma Awaz is Pakistan’s first bilingual AI screen reader, developed by Centangle Interactive to make digital content more accessible to blind and visually impaired users through Urdu and English text-to-speech. The platform addresses a major local-language accessibility gap where conventional screen readers have historically provided limited Urdu support.

The solution combines Urdu language processing, neural text-to-speech, natural voice output, offline voice models, Windows NVDA integration, and Android TalkBack support. Centangle worked across accessibility research, language processing, AI voice technology, software engineering, platform integration, testing, and public deployment to create an assistive technology solution designed around the needs of Pakistani users.

Key Challenges

The primary challenge was the lack of reliable, locally relevant assistive technology for Urdu-speaking blind and visually impaired users. Existing screen readers were largely optimized for English, resulting in poor Urdu pronunciation, limited local-language support, and barriers to accessing digital content independently.

The project also required the technology to work across different environments, including Windows desktop workflows and Android mobile devices. This meant solving challenges around Urdu text normalization, pronunciation, neural speech generation, natural voice quality, offline operation, accessibility integration, and low-latency performance.

Project Deliverables

Bilingual Urdu and English AI screen reader
Urdu text normalization and pronunciation engine
Neural text-to-speech pipeline
Natural male and female Urdu voices
Windows NVDA add-on
Android TalkBack integration
Offline and versioned voice models
Curated Urdu speech dataset
Accessibility-focused UX
Multi-platform accessibility workflows
Testing and speech validation
Public deployment and distribution

Project Solution

Centangle developed a bilingual AI-powered accessibility platform combining Urdu and English text-to-speech with natural voices and offline-first voice technology. The solution includes a dedicated Urdu text-processing engine, pronunciation rules, a neural TTS pipeline, downloadable voice models, and platform-specific integrations.

On Windows, Rehnuma Awaz is integrated through an NVDA add-on developed with Python. On Android, the solution works through native accessibility capabilities and TalkBack. The architecture also includes curated Urdu speech data, versioned offline voice models, local gRPC communication, and low-latency processing to provide responsive speech output without continuous internet dependency.

Project Outcome

Rehnuma Awaz created a locally relevant assistive technology solution for Urdu and English digital content, enabling blind and visually impaired users to access information through natural voice output across Windows and Android environments.

The platform has achieved more than 5,000 Android downloads and over 3,200 Windows public listing installs. It was designed to serve Pakistan's wider visually impaired community, addressing a population of more than 2 million blind and 24 million visually impaired people. The project demonstrates how AI, language technology, and accessibility engineering can be combined to reduce digital exclusion and improve digital independence.