A PDF is accessible when people can use it without seeing it or without a mouse: with a screen reader, a read-aloud function, magnification or a keyboard. For that, the PDF needs an invisible second layer that says what is a heading, a table, an image or a form field. That layer is usually missing – and that is exactly what DokAudit builds. The appearance of the document does not change.
This is what your PDF sounds like – before and after
Simplified examples, roughly as the widely used screen reader NVDA reads them out. On the left a typical PDF from a public administration, on the right the same passage after DokAudit.
Headings
Before – the screen reader reads: “Fee regulations of the municipal library Section 1 Scope These regulations apply to …” – all one single paragraph.
After DokAudit: “Heading level 1: Fee regulations of the municipal library. Heading level 2: Section 1, Scope.” The H key jumps from section to section.
What DokAudit does: Recognise headings from font size and formatting, tag them as H1 to H3 and generate bookmarks from them.
Tables
Before – the screen reader reads: “Service Fee Annual card 20 euros Reduced 10 euros Day pass 3 euros” – which number belongs to what?
After DokAudit: “Table with 2 columns and 4 rows. Service: Annual card. Fee: 20 euros.” The column heading is read out with every cell.
What DokAudit does: Tag tables as tables, define header cells and assign them to every cell; dissolve pure layout tables.
Images and charts
Before – the screen reader reads: “Graphic. Graphic. Graphic.” – or nothing at all; for the screen reader the chart does not exist.
After DokAudit: “Graphic: Visitor numbers of the municipal library 2022 to 2025, rising from 41,000 to 56,000.” The logo on every page is skipped.
What DokAudit does: Suggest alt text (AI, flagged for review), mark purely decorative images as decoration.
Forms
Before – the screen reader reads: “Edit, editable. Edit, editable. Check box, not checked.” – for what?
After DokAudit: “Surname, first name: edit. Date of birth: edit. I consent to the processing: check box, not checked.”
What DokAudit does: Give every field its visible label, hook the fields into the structure and order the tab sequence.
Two columns
Before – the screen reader reads: “The organic waste bin is emptied every Yellow bags are collected on the 14 days first Monday” – straight across both columns.
After DokAudit: “The organic waste bin is emptied every 14 days.” Then: “Yellow bags are collected on the first Monday.”
What DokAudit does: Set the reading order by columns and boxes – with AI support for difficult layouts, re-measured afterwards.
Headers and footers
Before – the screen reader reads: “… applies from Official Gazette No. 14 Page 3 of 12 1 January …” – in the middle of the sentence.
After DokAudit: “… applies from 1 January …” Repeated headers and page numbers are no longer read out.
What DokAudit does: Mark headers, footers, page numbers and dividing lines as artifacts.
Language and characters
Before – the screen reader reads: The German regulations are read with English pronunciation; “Gebühr” sounds like “Geb hr” or is missing entirely.
After DokAudit: A German voice; umlauts and special characters are pronounced correctly.
What DokAudit does: Set the document language, embed fonts and give every character its Unicode meaning.