Google Fonts Dataset
This dataset provides a structured collection of Google Fonts organized by script. Each script directory contains the fonts belonging to that script family.
The dataset is useful for:
- typography research
- multilingual OCR
- font classification
- document generation
- multilingual rendering pipelines
- language/script identification experiments
Dataset Structure
Fonts are organized by script code using ISO / Unicode script identifiers.
Example structure:
zipped_fonts/
├── Arab.zip/
│ ├── amiri/
│ ├── cairo/
│ ├── tajawal/
│ ├── scheherazadenew/
│ └── ...
├── Latn.zip/
├── Cyrl.zip/
├── Deva.zip/
├── Hani.zip/
├── Ethi.zip/
├── ...
Each script folder contains multiple font families, for example:
zipped_fonts/Arab.zip/
├── amiri
├── amiriquran
├── cairo
├── lateef
├── tajawal
└── vazirmatn
This allows easy filtering of fonts by writing system.
Scripts Included
The dataset includes fonts across a wide range of writing systems, including but not limited to:
- Arabic (
Arab) - Latin (
Latn) - Cyrillic (
Cyrl) - Devanagari (
Deva) - Greek (
Grek) - Hebrew (
Hebr) - Hangul (
Hang) - Japanese (
Jpan,Hira,Kana) - Chinese (
Hans,Hant,Hani) - Thai (
Thai) - Georgian (
Geor) - Armenian (
Armn) - Ethiopic (
Ethi) - Bengali (
Beng) - Tamil (
Taml) - Telugu (
Telu) - and many others.
The repository currently includes 100+ script directories.
Download
The recommended way to download the dataset is via git clone:
git clone https://huggingface.co/datasets/kargaranamir/google_fonts
Generating the Dataset
The dataset was created using the script:
get_fonts.py
This script collects and organizes Google Fonts into script-based directorie.
License
Fonts remain under their original licenses from Google Fonts.
Most fonts are licensed under one of:
- SIL Open Font License (OFL)
- Apache License 2.0
- Ubuntu Font License
Refer to the individual font directories for details.
Citation
This dataset was organized as part of the following project. If you use this dataset, please cite:
- Downloads last month
- -