Local startup ships Urdu speech-to-text engine for call centres

The model was trained on 40,000 hours of consented audio.

Officials said the development would bring direct benefits to citizens, with further measures expected in the coming weeks. Analysts broadly welcomed the move, though they cautioned that implementation would be the real test.

Background

Similar proposals were considered last year but never reached final approval. This year the landscape looks different, with all stakeholders appearing aligned on the way forward.

"The challenge is not the policy itself but sustaining it," one analyst noted.

Observers say the next few weeks will prove decisive. (This is a fictional demo story.)