GPU-Accelerated · Free · Open Source

Digitize Indic
Language Documents
with AI Precision

Upload scanned PDFs, images, or DOCX files in Tamil, Hindi, Telugu and 8 other Indic scripts. Get clean, downloadable text in seconds — powered by Surya Vision AI on cloud GPUs.

▶  Try It Free View on GitHub →
தமிழ் हिन्दी తెలుగు বাংলা ಕನ್ನಡ മലയാളം ગુજરાતી ਪੰਜਾਬੀ मराठी ଓଡ଼ିଆ English
11
Indic Languages
<7%
Character Error Rate
100
Max Pages per Run
T4
GPU Accelerated
$0
Cost While Idle

Everything you need to digitize
Indic language documents

Built on Vision Transformer models fine-tuned for complex Indic typography.

🔍
Layout-Aware OCR
Surya Vision AI detects text regions, headers, and columns before OCR — not just raw pixel scanning.
📄
Multi-Format Upload
Accepts PDF (up to 100 pages), PNG, JPG, TIFF, BMP, WEBP, and DOCX files in a single upload.
🌐
11 Indic Languages
Tamil, Hindi, Telugu, Bengali, Kannada, Malayalam, Gujarati, Marathi, Punjabi, Odia and English.
📊
Accuracy Metrics
Upload a ground-truth reference to get CER and WER scores for scientific evaluation of output quality.
⬇️
Instant Download
Download the extracted text as a clean UTF-8 .txt file with per-page headers for easy navigation.
Serverless GPU
Runs on Modal.com cloud T4 GPUs. Auto-scales to zero — $0 idle cost, near-instant processing.

Reads 11 languages natively

Including both historic and modern print styles across the Indian subcontinent.

தமிழ்
Tamil
South India
हिन्दी
Hindi
North India
తెలుగు
Telugu
Andhra / Telangana
বাংলা
Bengali
West Bengal / Bangladesh
ಕನ್ನಡ
Kannada
Karnataka
മലയാളം
Malayalam
Kerala
ગુજરાતી
Gujarati
Gujarat
मराठी
Marathi
Maharashtra
ਪੰਜਾਬੀ
Punjabi
Punjab
ଓଡ଼ିଆ
Odia
Odisha
Aa
English
Global

Up and running in 4 steps

No installation. No sign-up. Just upload and go.

Step 01
📁 Upload Your Document
Drop any scanned PDF (up to 100 pages), image file (PNG/JPG/TIFF), or DOCX into the uploader.
Step 02
🌐 Select Languages
Choose one or more of the 11 supported Indic languages from the checkbox selector.
Step 03
▶️ Run OCR
Click Run OCR. The Vision AI processes each page on a cloud T4 GPU and streams a live log.
Step 04
⬇️ Download Output
Download your clean UTF-8 text file. Optionally upload a reference to score CER and WER accuracy.

Try it right here

The full application is embedded below — no redirect needed.

Waking up GPU servers…
(~10 seconds on first load)