SemantiK
SemantiK turns scanned pages and PDFs into clean, accessible documents you can edit. SemantiK reads every page, works out the headings and structure, can describe pictures for people who use screen readers, and lets you check everything before you save.
Features
-
Scans to Structured Documents
Turn scanned pages, even ones that are only pictures of text, into documents with real headings, paragraphs, and pictures.
-
Models You Choose
Use the models you prefer, on your own computer or through a service you already use.
-
Review Before Export
See the result first, and fix any text, heading, or picture description before you save.
-
Accessible Exports
Save as Word, Markdown, plain text, an accessible web page, or an accessible PDF.
-
Desktop App and Command Line
Convert documents one at a time in the desktop app, or many at once from the command line.
-
Keyboard, Screen Reader, and Seven Languages
The desktop app works from the keyboard and with screen readers, in seven languages.
Supported Documents
SemantiK recognizes books, articles, reports, theses, manuals, letters, memos, emails, contracts, forms, slides, and newsletters, and handles anything else as a general document.
What SemantiK Opens
- Digital or scanned, including pages that are only pictures
- Images
- PNG, JPEG, TIFF, and WebP scans or photos of pages
- Web Pages
- HTML files
- Text
- Markdown and plain text files
What SemantiK Saves
- Word
- A .docx file with real headings and picture descriptions
- Accessible PDF
- A tagged PDF that screen readers can follow
- Accessible Web Page
- An HTML page with headings, contents, and picture descriptions
- Markdown
- Plain-text formatting for notes and websites
- Plain Text
- The words alone, without formatting
How SemantiK Works
-
Open a Document
Open a PDF, an image of a page, a web page, or a text file. Scans that are only pictures of pages work too.
-
SemantiK Reads the Pages
SemantiK reads the text, finds the headings, and can describe the pictures, showing progress as SemantiK goes. A long document can be paused and picked up later.
-
Check and Correct
Preview the result and fix any text, heading, or picture description. The original scan stays untouched.
-
Export
Save the document as Word, Markdown, plain text, an accessible web page, or an accessible PDF.
Before and After
Each pair shows an original page next to the same passage in the accessible web page SemantiK made from the page. The examples are a 1914 edition of A Christmas Carol, a NASA fact sheet, a report from the National Institute of Standards and Technology, and a 1911 issue of the newspaper Solidarity, all in the public domain. Scroll sideways in the preview, or focus the preview and use the arrow keys.
-
A Page of Text
Before: the scanned page, a picture of text that cannot be searched, resized, or read aloud. 
After: a real chapter heading and clean paragraphs, with the page number kept for reference. -
A Picture
Before: a color plate with no description, so a screen reader has nothing to say about the plate. 
After: the picture with the description SemantiK wrote, shown here as text; screen readers read the description aloud. -
A Photo
Before: a photo with only a short caption, so a screen reader cannot say what the photo shows. 
After: the photo with a description, opened here, of everything the photo shows. -
A Diagram
Before: a diagram that a screen reader can only skip over. 
After: the diagram with a full description, opened here, that traces every box and arrow. -
An Old Newspaper
Before: a blurred, speckled scan of 1911 newsprint with broken and faded letters. 
After: every word of the article as clean text that can be searched, resized, and read aloud.
SemantiK at Work
These captures come from real conversions in the SemantiK desktop app, of a 162-page scanned edition of A Christmas Carol from 1914 and a two-page contract. Scroll sideways in the preview, or focus the preview and use the arrow keys.
-
A conversion in progress: each stage shows how far along the stage is, and the whole run can be paused or stopped. -
Near the end of the run: the model's live text is shown as the model works out the chapter headings of the book. -
Checking the result: every change SemantiK made is listed and can be undone, next to the original page. -
A different kind of document: SemantiK recognizes a contract and marks its clauses, headings, and page headers. -
The converted contract as an accessible web page, ready to copy, open in a browser, or export.
Licensing
SemantiK is dual-licensed. You may use SemantiK under the GNU Affero General Public License, version 3 only (AGPL-3.0-only), or under a commercial license from Chiron Courseworks.
Frequently Asked Questions
- Can I use SemantiK without paying?
- Yes. The AGPL-3.0-only license lets you use, study, change, and share SemantiK. In return, if you distribute SemantiK, or let others use a modified version over a network, you must make the corresponding source code available under the same license.
- When do I need a commercial license?
- When those conditions do not work for you: for example, building SemantiK into a product or service whose source you will not release, or working in an organization whose policies do not permit AGPL software. A commercial license releases you from the AGPL obligations.
- How is a commercial license priced?
- Commercial licenses are quoted for each organization. Tell us your organization's size and how you plan to use SemantiK, and we will reply with terms.
- How do I get SemantiK?
- SemantiK is released on GitHub at github.com/mdmurphy822/SemantiK, the official source for releases and source code. For a commercial license, contact us.
- Which models does SemantiK use?
- The models you choose, on your own computer or through a service you already use. A setup guide in the desktop app helps you connect them. The models are not bundled with SemantiK.
- Does SemantiK need an internet connection?
- No. With models on your own computer, SemantiK runs entirely offline, and your documents never leave your computer.