DEV Community: Thomas Scharl

Extract Text from Any Scanned Document for Free: Client-Side OCR with Tesseract.js & PDF.js

Thomas Scharl — Tue, 10 Nov 2026 12:34:33 +0000

Extract Text from Any Scanned Document for Free: Client-Side OCR with Tesseract.js & PDF.js

Have you ever needed to copy text from an invoice, a photo of a book page, or a scanned contract, only to discover it’s saved as a flattened image without a searchable text layer?

Usually, your options are:

Re-type the entire document by hand.
Pay for Adobe Acrobat Pro or an enterprise OCR subscription.
Upload your private financial receipts to a random free OCR website that stores your data on unverified servers.

In this guide, we'll demonstrate how modern browser technology allows you to perform high-accuracy OCR directly on your client machine using WebAssembly, ensuring that sensitive data never leaves your device.

We have integrated this exact workflow into TnTS-Tools OCR, a 100% free, browser-based tool suite.

Why Traditional Cloud OCR is a Privacy Hazard

When you upload a scanned invoice or tax declaration to a cloud-based OCR service:

Your files travel across the public internet.
Many free web tools retain uploaded images to train AI models or monetize user data.
For business users and EU freelancers, this violates GDPR data processing requirements without a signed DPA.

By shifting optical character recognition to WebAssembly running inside the browser, your CPU handles the matrix calculations. The network payload is literally 0 bytes.

How It Works Under the Hood

1. Rendering Scanned PDF Pages with `pdf.js`

Scanned PDFs don't contain raw image streams in an easily copyable format. Mozilla’s pdf.js library allows us to render any PDF page onto an invisible HTML5 <canvas> element at 300 DPI for optimal recognition accuracy.

2. Neural OCR Extraction with `Tesseract.js`

The canvas bitmap is passed to Tesseract.js, a WebAssembly port of the renowned open-source Tesseract OCR engine.

Multi-language support (German, English, French, Spanish, Italian).
Multi-threaded Web Worker execution keeps the UI smooth and responsive.
Character confidence scoring ensures high recognition accuracy.

3. Instant Formatting & One-Click Export

Once extracted, text is cleaned, paragraphs are structured, and users can copy the text to their clipboard or download a clean .txt document with a single click.

Try It Online (No Sign-Up Required)

If you have a scanned document, receipt, or screenshot you need to convert to editable text right now:

👉 Launch TnTS-Tools OCR Texterkenner

100% Free (No subscription, no credits)
100% Private (Files stay in your browser memory)
Zero Registration (No email, no accounts)

Feedback and feature requests are always welcome!

Say Goodbye to Monthly PDF Subscriptions: How Modern Browser Tech is Replacing Paid SaaS

Thomas Scharl — Tue, 10 Nov 2026 12:31:52 +0000

Say Goodbye to Monthly PDF Subscriptions: How Modern Browser Tech is Replacing Paid SaaS

How many times have you reached for an online tool just to do something trivial—like merging two PDF invoices, extracting pages, or reducing file size for an email—only to be slammed by:

"You have reached your daily limit of 2 tasks"
"Sign up with Google to continue"
"Enter your credit card for a 7-day trial ($18.99/mo after)"
Or worse: An invisible upload of your sensitive tax records or NDA-protected contracts to a server somewhere on the other side of the planet.

It’s 2026. Your browser is essentially a full-fledged operating system capable of running WebAssembly, WebGL, and multi-threaded Web Workers at near-native speeds.

There is zero architectural reason why a simple PDF merge or compression needs to leave your local device.

In this post, we’ll look at why we built TnTS-Tools — a completely free, 100% client-side web toolkit for everyday document and image workflows — and how modern browser APIs make traditional subscription-heavy file utilities obsolete.

The Dirty Secret of Document Utilities

Most popular PDF SaaS platforms operate on a legacy cloud model:

You select a 20 MB PDF.
The browser uploads the entire 20 MB over your internet connection to their server cluster.
Their backend spins up headless Python or C++ binaries (like ghostscript or poppler), manipulates the bytes, and writes a new file to cloud storage.
You wait to download the result.

This approach has three massive downsides:

Bandwidth & Latency: Uploading and downloading huge files takes time, especially on mobile or slower connections.
Privacy & Compliance Nightmares: Under GDPR and privacy laws, uploading employee contracts, medical records, or proprietary blueprints to third-party cloud servers requires strict Data Processing Agreements (DPAs).
Enormous Server Bills = Aggressive Paywalls: Cloud compute and egress bandwidth cost money. That’s why these platforms force users into expensive monthly subscriptions.

Moving the Compute to Where the Data Lives: The Browser

Modern web technology has fundamentally shifted what is possible inside a standard HTML5 tab:

1. In-Memory PDF Manipulation with `pdf-lib`

Instead of sending files to a backend, libraries like pdf-lib operate purely in-memory using JavaScript typed arrays (Uint8Array / ArrayBuffer).

When you drag and drop five PDF documents into TnTS-Tools, the JavaScript engine parses the cross-reference tables and streams directly on your CPU. Pages are extracted and merged into a new document in milliseconds.

Network payload? 0 bytes. Your files never touch a server.

2. Client-Side Rendering with `pdf.js`

Need to preview pages or reorder them before merging? Mozilla’s pdf.js compiles PDF view logic into canvas drawing calls. Users can drag thumbnails, rotate orientation, or delete blank pages with zero round trips.

3. Edge Text-to-Speech (Neural Voice Reader)

One of the coolest features we added to TnTS-Tools PDF Suite is an interactive document reader. Using Edge Neural TTS integration, lengthy PDF reports, academic papers, or ebooks can be listened to in studio-quality neural voices across 8 languages—sentence by sentence with real-time text highlighting.

No Signups. No Subscriptions. Just Tools That Work.

We built TnTS-Tools based on three strict rules:

Zero Registration: No accounts, no password fatigue, no marketing spam.
Privacy by Design: Everything that can run locally, runs locally.
100% Free Access: Everyday productivity should not cost $15 every month.

If you are tired of subscription walls for everyday document tasks, feel free to try it out:
👉 https://tntstools.com

What are your favorite client-side web tools? Let's discuss in the comments below!

Why We Built TnTS Tools: A 100% Free, Client-Side PDF Suite & Neural AI Voice Reader

Thomas Scharl — Tue, 10 Nov 2026 08:20:31 +0000

Why We Built TnTS Tools: A 100% Free, Client-Side PDF Suite & Neural AI Voice Reader

Have you ever tried to do something trivial in your browser—like compress a 12 MB PDF for an email, remove a background from a photo, or extract text from a scanned document?

If you have, you know how frustrating the modern web has become:

"You have reached your 2 free tasks for today."
"Create an account and start your 7-day free trial ($18.99/mo)."
Mandatory email signups, paywalls, and uploading your personal or business documents to unknown ad-tracking servers.

We were tired of this artificial friction. That is why we built TnTS Tools (and our Austrian mirror ttstools.at)—a completely free, privacy-first web suite designed to solve daily productivity tasks right inside your browser without accounts, subscriptions, or limits.

🛠️ What We Built (And How It Works Under the Hood)

1. In-Browser PDF Studio (`pdf.php`)

Most PDF web services upload your entire document to their backend servers to run headless LibreOffice or ghostscript.

Instead, we chose a client-side first architecture:

Merging, Splitting & Compression: Powered by pdf-lib running directly inside the user's browser thread. Your sensitive contracts, invoices, and resumes never leave your device.
Visual Page Reordering: We use pdf.js to render crisp, interactive page thumbnails on the fly. Users can drag, drop, swap, or delete individual pages visually before generating the final document.

2. Neural Text-to-Speech (TTS) Voice Reader

One of our favorite features is the PDF & Text Voice Reader.

Instead of robotic, monotone web-speech synthesisers, we set up a dedicated microservice running on FastAPI and neural models.

Natural Studio Voices: Features human-sounding voices across multiple languages and dialects:
- 🇩🇪 German: Conrad (natural male), Katja (clear female), Killian (warm)
- 🇦🇹 Austrian: Ingrid (female), Jonas (male)
- 🇨🇭 Swiss German: Leni
- 🇺🇸 🇬🇧 English: Jenny, Guy, Sonia
- 🇫🇷 🇪🇸 🇮🇹 French, Spanish, Italian
Synchronized Sentence Highlighting: As the voice speaks, the text automatically scrolls and highlights the active sentence in real time—ideal for students studying long lecture notes, proofreading, commuters, or users with dyslexia.
Built-in OCR Scan Fallback: If you drop a scanned image PDF without an embedded text layer, the reader automatically detects it and offers 1-click in-browser OCR extraction via Tesseract WebAssembly.

3. AI Image Magic: 1-Click Segmentation & Generative Inpainting

For bild.php, we wanted a seamless experience:

Interactive 1-Click Object Selection: Using Meta's Segment Anything (SAM) model, users simply click on a person, object, or watermark. The model accurately detects the boundary of the entire object.
Generative Erasing: Instead of simple blur filters or smudges, generative inpainting seamlessly reconstructs the background textures (walls, furniture, natural lighting) as if the object was never there.
Alpha Channel Background Removal: Clean cutouts exported directly as transparent PNGs or WebP.

4. Client-Side WASM OCR (`ocr.php`)

For digitizing receipts, book scans, and documents:

Fully client-side via Tesseract.js compiled to WebAssembly.
Extracts clean text in seconds with 1-click clipboard copying and TXT download.
No data transmitted over the wire—maximum GDPR and data privacy compliance.

🚀 Try It Out & Give Us Your Feedback!

TnTS Tools is live and completely free for everyone:
👉 https://tntstools.com

We would love to get your thoughts:

What daily file utility or tool are you currently missing in your workflow?
How is the playback speed and voice naturalness on the TTS player?
Any UI/UX tweaks you'd recommend?

Drop your feedback in the comments below! 👇

DEV Community: Thomas Scharl

Extract Text from Any Scanned Document for Free: Client-Side OCR with Tesseract.js & PDF.js

Extract Text from Any Scanned Document for Free: Client-Side OCR with Tesseract.js & PDF.js

Why Traditional Cloud OCR is a Privacy Hazard

How It Works Under the Hood

1. Rendering Scanned PDF Pages with pdf.js

2. Neural OCR Extraction with Tesseract.js

3. Instant Formatting & One-Click Export

Try It Online (No Sign-Up Required)

Say Goodbye to Monthly PDF Subscriptions: How Modern Browser Tech is Replacing Paid SaaS

Say Goodbye to Monthly PDF Subscriptions: How Modern Browser Tech is Replacing Paid SaaS

The Dirty Secret of Document Utilities

Moving the Compute to Where the Data Lives: The Browser

1. In-Memory PDF Manipulation with pdf-lib

2. Client-Side Rendering with pdf.js

3. Edge Text-to-Speech (Neural Voice Reader)

No Signups. No Subscriptions. Just Tools That Work.

Why We Built TnTS Tools: A 100% Free, Client-Side PDF Suite & Neural AI Voice Reader

Why We Built TnTS Tools: A 100% Free, Client-Side PDF Suite & Neural AI Voice Reader

🛠️ What We Built (And How It Works Under the Hood)

1. In-Browser PDF Studio (pdf.php)

2. Neural Text-to-Speech (TTS) Voice Reader

3. AI Image Magic: 1-Click Segmentation & Generative Inpainting

4. Client-Side WASM OCR (ocr.php)

🚀 Try It Out & Give Us Your Feedback!

1. Rendering Scanned PDF Pages with `pdf.js`

2. Neural OCR Extraction with `Tesseract.js`

1. In-Memory PDF Manipulation with `pdf-lib`

2. Client-Side Rendering with `pdf.js`

1. In-Browser PDF Studio (`pdf.php`)

4. Client-Side WASM OCR (`ocr.php`)