How LeedyBox Works

Browser-Based File Processing

LeedyBox tools run as JavaScript in your web browser. When you choose a file or enter text, the browser reads it on your device and the processing runs on your computer or phone. Files and text are not uploaded to a LeedyBox server for processing. Many tools use standard browser features: the Canvas API for images, the Web Crypto API for hashes, passwords and UUIDs, Intl.Segmenter for word counts and the Intl date APIs for time zones. The OCR in PDF to Text and Image to Text uses Tesseract.js, which runs WebAssembly code inside the browser.

What Happens to Your Files?

1. You select or drag and drop a file from your device, or type or paste text into the tool.
2. The browser reads the file into memory for the current page.
3. The tool converts, edits, combines, splits or reads the file with JavaScript libraries running on your device.
4. The result is created in the browser and shown on the page or offered as a download. How long this takes depends on the input and your device.
5. A copy button writes the result to your clipboard only when you click it.
6. Closing or reloading the page discards the files, text and generated values held in that page's memory.

Libraries Used by Each Tool

Some tools are built on open-source libraries; others use only LeedyBox's own scripts and features built into your browser. Every library, together with the data it needs (font tables, OCR language data, models), is served from leedybox.com itself at a fixed version; tool pages do not load code from public content delivery networks (CDNs). Only the library code and data are requested; the files and text you process are not part of these requests.

The tool pages also load the Tailwind CSS script (version 3.4.17) from leedybox.com for styling. The license texts of the libraries are in LICENSES.txt.

Why Local Processing Matters

Processing on your device means a file does not have to travel to a conversion server and wait in a queue, and LeedyBox never receives a copy of it. It also means the result depends on your device: a computer with plenty of memory can handle larger files than an older phone.

Browser Limitations

Because processing relies on your device's resources, performance depends on available memory, browser capabilities, CPU speed and file complexity. Very large or complex files need more memory and time. If a tab runs out of memory, the browser may stop the page; splitting the work into smaller files usually helps. Some output formats also depend on what your browser can encode.

Privacy and Network Requests

Local processing keeps your files on your device, but the web page itself still uses the network: it loads its own HTML, styles and scripts and the libraries listed above, all from leedybox.com, and it loads Google Analytics and Google AdSense, which store cookies only if you accept them. Cloudflare, which hosts the site, also measures how quickly pages load (Cloudflare Web Analytics, without cookies). None of these requests include the contents of the files or text you process, or the passwords, UUIDs, hashes and codes you generate.

PDF to Text and Image to Text download the OCR engine and language data from leedybox.com when they first need them. Tesseract.js keeps the downloaded language data in your browser's IndexedDB storage so that it does not have to be downloaded again; you can delete it by clearing this site's data in your browser.

The ID Photo Maker downloads the MediaPipe library, its WebAssembly runtime and a face model from leedybox.com when you choose the first photo: about 7 MB to download (about 15 MB once unpacked). A second, small model (0.25 MB) is downloaded only if you replace the background. They are kept only in the browser’s normal cache; your photo is not sent anywhere.

The Simplified ↔ Traditional Chinese converter downloads opencc-js and the dictionaries of the chosen conversion from leedybox.com when you start using it: up to about 100 KB from Traditional, about 1.1 MB to Traditional. Your text is not sent anywhere.

Audio to Text downloads Transformers.js, the ONNX Runtime WebAssembly engine and the Whisper model you choose from leedybox.com when you start the first transcription: about 90 MB for base, 250 MB for small or 750 MB for large-v3-turbo, which runs on the graphics processor through WebGPU. They are saved in the browser’s Cache Storage so that they are not downloaded again; the tool shows their size and can delete them. The part of the page that runs the model is not allowed to contact any other site, and your recording is not sent anywhere.

Related: Read Privacy Policy · Read Cookie Policy

Frequently Asked Questions