---
pretty_name: Chagatai OCR
language:
  - chg
  - kk
tags:
  - ocr
  - chagatai
  - transcription
size_categories:
  - n<1K
extra_gated_heading: Request access
extra_gated_description: Access is reviewed and approved manually by the dataset owner.
extra_gated_button_content: Request access
---

# Chagatai OCR

Chagatai text in Arabic script, with Cyrillic transcriptions and Kazakh translations.

The dataset is available as `Dataset_OCR.xlsx` and UTF-8 `Dataset_OCR.csv`. Both contain 875 non-empty data rows.

| Column | Contents |
| --- | --- |
| `Original` | Text in Arabic script |
| `Transcript` | Cyrillic transcription |
| `Translation (?)` | Kazakh translation; the original column label is retained |
| `Pages` | Page markers, provided on selected rows |

The Excel workbook is provided unchanged. The CSV contains the same values from its only worksheet (`Лист1`). Some fields are empty. Page images and train/test splits are not included.

## Access

Request access on this page. Downloads become available after the dataset owner approves your request.

Once approved, sign in and download the workbook:

```bash
hf auth login
hf download chagatai-project/ocr --repo-type dataset --local-dir ./ocr
```