Edit & correct recognized text
Search, review low-confidence regions, edit the text layer, and let Scanix auto-save your corrections.
OCR is rarely perfect on difficult scans, so every word Scanix reads stays editable. After a page is read by OCR, open the Extracted Text panel in the Viewer to search the recognized text, step through the regions Scanix flagged as uncertain, and fix any wording — and by default each correction saves the moment you make it. This page walks through that review-and-fix workflow end to end.
Corrections rebuild the searchable layer
Fixing text here doesn't just tidy the on-screen list. Your edits are written back into the invisible, searchable text layer that ends up in the exported Searchable PDF, so a correction you make now is a correction anyone searching the final file benefits from.
Open the Extracted Text panel
The panel lives in the Viewer's right sidebar, on the Text tab. It is headed Extracted Text with a count of how many regions are on the current page — for example Extracted Text (24). Each region is its own card: an editable label at the top, the recognized text in a box you can type into, and a status line showing how confident OCR was.
For a document opened without a template (a free-form import), the recognized text appears under the single Fields tab instead of a separate Text tab — the panel and its rows work exactly the same.
Screenshot
The Viewer right sidebar on the Text tab, showing the Extracted Text panel header with a region count, the search bar below it, and a stack of OCR region cards. — shot ocr-editing-and-correcting-01
Find the region you need
Search the recognized text
Type into the Search in OCR text... box at the top of the panel. The list filters down to the matching regions and shows a count — "N regions match" — so you can jump straight to the word you're looking for instead of scrolling. Clear the box (the ✕ inside it) to see every region again.
Review the low-confidence regions
Next to the panel title is a button showing how many regions Scanix is unsure about — labelled N issues (its tooltip reads Review low-confidence regions). Click it to switch the list to review mode, which hides the regions OCR is confident about and leaves only the ones worth a second look.
In review mode a small navigator appears with up and down arrows and a position / total counter, so you can walk through each flagged region one at a time. As you move, the matching box highlights on the page image, so you always see exactly what text the region covers.
Select a region on the page or in the list
Click any region card to select it, or click its box on the page image — the two stay in sync. The selected card scrolls into view and its box is highlighted, which is the fastest way to line up the recognized text against the original scan before you edit.
Fix the text
Once you've found a region, every part of its card is editable.
Retype or correct the text
Click into the text box on the region card and edit it like any text field. It starts as a single line to keep the list compact, but you can drag its bottom edge to expand it when a region spans several lines — a multi-line address or a paragraph, for example. Once you change a region, an Edited badge appears on its status line so you can see at a glance which text you've touched.
Rename a region (optional)
The label at the top of each card is editable — click it and type to give the region a meaningful name. Leaving it empty restores the automatic default (such as Region #3). This is handy when you want a captured region to read clearly rather than as a generic number.
Change a region's type (optional)
Each card has a small type dropdown (for example text, title, table, date, or barcode). Pick the type that matches what the region actually is when the automatic guess is wrong — it tells Scanix how to treat that region's content.
Delete a region
If OCR picked up something that isn't really text — a logo, a stray mark, a duplicate box — click the trash icon (Delete region) on that card to remove it.
What the status line tells you
Each card shows an OCR N% confidence reading. When Scanix is least sure, it changes to Low OCR confidence (N%) with a warning icon — a cue that the region is a good candidate to verify against the scan. The exact text you read is always the text that gets saved, regardless of the confidence number.
Screenshot
A single OCR region card with its editable label, the recognized text in an expandable box, the region type dropdown and a delete button in the header, and an OCR N% status line below. — shot ocr-editing-and-correcting-02
Saving your corrections
By default, Scanix saves your text corrections the instant you make them — there's nothing to remember to click.
Auto-save (the default)
The behaviour is controlled by a single global preference: Settings → General → Auto-save OCR corrections (recommended), which is on out of the box. While it's on:
- Edits to the recognized text layer — and moving or resizing a region's box — persist instantly.
- There is no Save button. In its place a small status line keeps you informed: it reads Saving… while a change is being written, flashes Saved when it lands, and otherwise shows Auto-save on.
Screenshot
The Settings → General page with the Auto-save OCR corrections (recommended) checkbox switched on. — shot ocr-editing-and-correcting-03
Manual save (turn auto-save off)
If you'd rather commit corrections deliberately — common for QA and verification operators — turn Auto-save OCR corrections (recommended) off in Settings → General. A Save button then reappears in the Viewer sidebar. It lights up only when you have unsaved text changes, and you can trigger it with Ctrl+S. Until you press it, your text edits aren't committed.
Save governs the OCR text layer — index fields always save on their own
The Save button (and Ctrl+S) commits OCR text-layer and region corrections only. Index field values — the invoice number, the date, anything captured by a template's fields — save automatically the moment you commit them, no matter how the Auto-save OCR corrections preference is set. So turning auto-save off never risks losing a field value; it only adds a deliberate commit step for the recognized-text layer.
Troubleshooting
Next steps
Run OCR
Read pages on demand or automatically after import, and re-run a page when needed.
Learning from your corrections
How the fixes you make help Scanix get the next document right.
General settings
Where the Auto-save OCR corrections preference lives, alongside the other app defaults.