Zuhri is a comprehensive and manually verified Urdu Grapheme-to-Phoneme (G2P) dataset. It is designed to aid research and development in areas such as speech synthesis, pronunciation modeling, and computational linguistics, specifically for the Urdu language.
This dataset provides accurate phoneme transcriptions and IPA representations, making it ideal for use in building high-quality TTS (Text-to-Speech), ASR (Automatic Speech Recognition), and other phonetics-related systems.
Grapheme-to-Phoneme (G2P) conversion is the task of translating written text (graphemes) into spoken sounds (phonemes). G2P is crucial for generating natural-sounding speech from text and for correct pronunciation modeling in voice-driven technologies.
The dataset is stored in the data/ directory as a CSV file. Each row in the file contains the following five columns:
Word, Phoneme Sequence (space-separated), IPA Transcription, Verified (True/False), Frequency Rank
| Word | Phonemes | IPA | Verified | Rank |
|---|---|---|---|---|
| عائشہ | ʕ aː ɪ ʃ ə | /ʕaːˈɪʃə/ | True | 1 |
| عطیہ | ʕ ʈ iː j ə | /ʕʈiːˈjə/ | True | 1 |
| طور | t̪ uː r | /t̪uːr/ | True | 1 |
| اردو | ʊ r d uː | /ʊrˈduː/ | True | 1 |
| اللہ | ɑ l l aː h | /ɐlˈlɑːh/ | True | 1 |
This dataset is useful for:
- Urdu Text-to-Speech (TTS) systems
- Automatic Speech Recognition (ASR)
- Phonetic alignment and transcription
- Linguistic and phonological research
- Speech technology applications in Urdu
git clone https://github.com/humair-m/zuhri.git
cd zuhriimport pandas as pd
# Load the Urdu G2P dataset
df = pd.read_csv("data/urdu_g2p.csv")
# Display the first few rows
print(df.head())If you use this dataset in your research, please cite it as follows:
Humair Munir. URDU G2P: Zuhri Urdu Grapheme-to-Phoneme Dataset. Version 1.0, 2025. Licensed under Apache 2.0.
@humair-pc{humair2025urdug2p,
author = {Humair Munir},
title = {{URDU G2P: Zuhri Urdu Grapheme-to-Phoneme Dataset}},
year = {2025},
version = {1.0},
howpublished = {\url{https://github.com/humair-m/zuhri}},
note = {Apache License 2.0}
}📧 For academic queries, contact: humairmunirawan@gmail.com
This dataset is released under the Apache License 2.0. You are free to use, modify, and distribute it under the terms of this license. See the LICENSE file for full details.
We welcome contributions from the community! You can:
- Suggest corrections or improvements in phoneme alignments
- Add more verified Urdu words
- Extend the dataset for regional dialects
- Submit pull requests for related G2P tools or training scripts
Please open an issue to discuss your suggestions before submitting a PR.
For questions, support, or collaborations, feel free to reach out:
📧 Email: humairmunirawan@gmail.com
This dataset is part of an ongoing effort to support low-resource languages like Urdu in the field of speech and language technologies. Thank you to the open-source community for continuous support and feedback.