Publications/Natural Language Processing & LLMs for Low-Resource Languages
Conference · 2018
PronouncUR: An Urdu Pronunciation Lexicon Generator
Haris Bin Zia, Agha Ali Raza, Awais Athar
LREC 2018, Miyazaki, Japan
Summary
A speech recognizer needs a pronunciation lexicon, a list telling it how each word sounds, and building one normally takes expert linguists. PronouncUR is a tool that generates pronunciations for any list of Urdu words. It uses an LSTM-based model trained on a handcrafted expert lexicon of about 39,000 words and predicts pronunciations with 64% accuracy. More importantly, when its output is used inside a speech recognition system, the word error rate is comparable to that obtained with a fully handcrafted expert lexicon.
Cite this paper
Haris Bin Zia, Agha Ali Raza, Awais Athar. PronouncUR: An Urdu Pronunciation Lexicon Generator. LREC 2018, Miyazaki, Japan.
@inproceedings{zia2018pronouncur,
author = {Haris Bin Zia and Agha Ali Raza and Awais Athar},
title = {{PronouncUR: An Urdu Pronunciation Lexicon Generator}},
booktitle = {LREC 2018, Miyazaki, Japan},
year = {2018},
url = {https://aghaaliraza.com/publications/2018--LREC--PronouncUR_An_Urdu_Pronunciation_Lexicon_Generator.pdf}
}
- IRTIQA: Evolving in English, Evaluated in Urdu — Does Agent Self-Evolution Transfer Across Languages?Findings of EACL 2027, Athens, Greece
- UrduCultureQA: A Culturally Grounded Benchmark for Urdu Question Answering and Alignment EvaluationFindings of EACL 2027, Athens, Greece
- Pak3H: Evaluating the Cost of Cultural Mismatch in LLM Alignment with a Human-Contextualized Urdu BenchmarkEMNLP 2026 (Main), Budapest, Hungary
- Evaluating Robustness of LLM-Based Ambient Scribes for SOAP Note GenerationCHIL 2026 (Conference on Health, Inference, and Learning), PMLR vol. 333
- Kahaani: A Multimodal Co-Creative Storytelling SystemEACL 2026 Student Research Workshop, Rabat, Morocco