Scan to Text: Paper to Words You Can Paste Anywhere
By Schema LawPublished Updated
Scan to text means two moves, not one. Scan the paper page to a PDF, then run OCR on that scan and export the result as plain text. The text is ordinary characters, so you can paste it into an email, a case note or a form field, once you have checked it against the page.
Nobody wants a text file for its own sake. You want the words from a letter in a file note, a paragraph from a lease in an email to the client, or a description from a police report in a court's online form. Where the words are going decides how much of the page to convert and what cleanup they need before you press send.
What does scan to text give you that scan to PDF does not?
Scan to text gives you characters. Scan to PDF gives you a picture of the page. Characters can be pasted, searched, counted, trimmed and typed over. A picture can only be looked at, attached or printed. The page content is the same; what your computer can do with it is not.
Set side by side:
| You want to | Scan to PDF (picture) | Scan to text (characters) |
|---|---|---|
| Show someone exactly what the page looked like | Yes | No. Letterhead, signatures and stamps are not carried |
| Paste a sentence into an email body | No | Yes |
| Put a passage in a web form | No. A form field takes typing, not a picture | Yes |
| Search for a name | No | Yes |
| Count the characters in a paragraph | No | Yes |
| Prove what the paper said | Yes. It is the record | No. It is a reading of the record |
The last row is why scan to text never replaces the scan. The text is produced by software reading the picture, and software misreads. Keep the PDF as the record and treat the text as a working copy taken from it.
What "text" means here
Plain text is the barest form: letters, digits, punctuation, spaces and line endings. It has no fonts, no bold, no margins and no tables. That bareness is the point when the destination has its own formatting or accepts none. Word, Markdown and HTML exports carry more structure, and each suits a different destination. If you want the page back as a document you can edit and print, the route is different, and the post on scanning text to Word covers it from the scanner glass onwards.
Scanning a paper page and lifting the text out of the scan
Scan the page to a PDF at settings that keep small print sharp, look at the scan before the paper leaves your desk, run OCR on it, export plain text, and read the text against the scan. Then copy what you need. Six steps, and the fifth is the one that protects you.
A worked example. Priya, a paralegal, has a one-page letter from a fictional landlord, Calder Row Estates, to the firm's client, Jonah Whitlock, giving notice of a rent increase. The supervising lawyer wants the operative paragraph in the file note and in an email to the client.
- Scan the page to a PDF. Priya uses the copier's text-document preset and sends the file to her own folder. She names it
Whitlock - Calder Row rent notice 2026-09-30 (scan).pdf. - Look at the scan. One page, upright, nothing cut off at the edges, the smallest print readable when she zooms in. The letter goes back in the file only after this check.
- Run OCR. She uploads the PDF to the firm's OCR tool. In Schema Law that is a browser upload of a PDF, PNG or JPEG of up to 20 MB, after which the page is read and shown on screen. The three steps on the homepage describe the flow.
- Export plain text. She downloads the
.txtfile and saves it beside the scan with a matching name ending(OCR text).txt. - Read the text against the scan. With both open, she checks the new rent figure, the effective date, the address of the premises and the landlord's name. The date in the text reads "1 Novernber 2026". She corrects it in the text file and notes the correction.
- Copy the paragraph. She opens the text file in a plain editor, selects the operative paragraph and copies it.
Open the text in a plain editor first
Step six matters more than it looks. A plain editor such as Notepad on Windows or TextEdit in plain-text mode on a Mac shows exactly the characters that will be pasted, with nothing added. If you open the .txt file in a word processor and copy from there, the clipboard can carry a font and paragraph settings along with the words, and the destination may honour them.
When the scanner offers "scan to text" itself
If your copier or scanning program has a button that produces a text file directly, the recognition still happens. It happens inside the device or its software. If you use that route, make sure the device also saves the scan. A text file with no picture behind it cannot be checked once the paper is gone.
Where is the scanned text going: an email, a case note or a form field?
The destination decides the cleanup. An email wants readable paragraphs and a note of where the words came from. A case note wants the passage, the source and the date it was checked. A form field wants plain characters inside a length limit. Settle the destination before you copy, and you clean the text once.
| Destination | What it accepts | Cleanup before pasting | What to add |
|---|---|---|---|
| Email body | Paragraphs; formatting depends on the mail program and the recipient's | Join broken lines; remove page headers and footers caught in the text | Quotation marks or an indented block, and the document's name and date |
| Case note or file note | Plain paragraphs in your practice system's note field | Join broken lines; keep the wording exact | Source file name, page, who checked it and when |
| Form field | Plain characters, sometimes with a limit and a restricted character set | Remove line breaks; replace special characters the form rejects; trim to the limit | Nothing inside the field; record the source in your own file |
| Chat or text message | Short plain text | Shorten to what the reader needs | A line saying it is an extract and the full document follows |
Paste as plain text, then format the quotation yourself. If the paragraph arrives chopped into short lines, each ending where the printed line ended, the breaks need joining before the email is readable on a phone. The cause and the repair are in the post on copying from scans, under why text pasted from a scanned PDF breaks at the end of every line.
Plain text cannot carry a table. If the passage you are sending includes a rent schedule or a list of payment dates in columns, the columns collapse into a run of words and figures. For that case an HTML export keeps the structure, and the post on scanned PDF to HTML explains pasting a converted scan into an email with its tables intact.
Say what the reader is looking at. "The notice says:" followed by a quoted paragraph tells the client these are the landlord's words. Attach the scan as well when the reader may need to see the page.
That notice is often a landlord writing to a tenant under a lease. Keep the premises, the date and the section the notice cites. If it mentions quiet enjoyment or the security deposit, quote that sentence as printed and name the scan page in the note.
Case note
A note field in a practice management system is a text box with a history. Whatever you paste will be read months later by someone who has not seen the paper. Paste the passage exactly, without tidying the landlord's grammar, and put your own summary in a separate paragraph so nobody mistakes one for the other.
Form field
Online forms are the least forgiving destination, because they can reject or silently alter what you paste. The general rule: paste into a plain editor first, fix the text there, then paste into the form.
Fitting scanned text into a form field with a character limit
To fit scanned text into a limited form field, find out the limit and how it counts, strip the text to plain characters, count it in an editor, and trim it without changing its meaning. Do all of that outside the form, then paste once and read the field back before you submit.
Find out how the form counts
Limits differ from one form to the next, and so does the way each counts. Look for a counter beside the field or a stated maximum in the form's instructions. Three things to find out by testing with harmless text:
- Whether spaces count toward the limit.
- Whether a line break counts as one character, two or none.
- What happens at the limit: the form stops accepting input, cuts the end off without telling you, or shows an error on submission.
A form that cuts text off silently is the dangerous kind. A description that ends mid-sentence, or loses its final "not", is worse than a shorter one written to fit.
Reduce the text to plain characters
Scanned text can carry characters a form will not take. Replace them in your editor before pasting.
| In the scanned text | Why it causes trouble | Replace with |
|---|---|---|
| Curly quotation marks and apostrophes | A form may store them as question marks or boxes | Straight quotes |
| Long dashes | Same problem | A hyphen, or a comma where the sense allows |
| Section and paragraph signs | May be rejected or altered | The word "section" or "paragraph" |
| Joined letters such as a single "fi" character | Looks right, fails a search later | The two separate letters |
| Tabs and repeated spaces | Collapse unpredictably | A single space |
| Line breaks inside a paragraph | May count against the limit or be removed with no space left behind | A single space |
After pasting, read the field itself. If a character has turned into a question mark, the form did not accept it, and what you see in the field is what will be filed.
Count before you paste
Count in the editor, not by eye. A word processor's word count tool reports characters with and without spaces for whatever you have selected. Compare the figure with the limit, and leave a small margin in case the form counts line breaks differently from your editor.
Trim without changing the meaning
Suppose a court's online intake form has a 500-character field for "description of the notice received", and the operative paragraph of the Calder Row letter runs to 742 characters. Priya has three honest options.
- Quote the core and say it is an extract. Keep the sentence that states the new rent and the date, in quotation marks, and mark any omission inside it in the way your court's or office's citation style requires.
- Summarize in your own words. "Letter dated 30 September 2026 from Calder Row Estates giving notice that monthly rent rises to $2,150 from 1 November 2026." No quotation marks, because it is no longer the landlord's wording.
- Refer to an attachment. Where the form allows uploads, give a one-line description and attach the scan.
What she must not do is delete words from the middle of a quoted sentence until it fits and leave the quotation marks on. Whether a field calls for the exact words or a summary is a question for the form's instructions and the supervising lawyer.
One paragraph or the whole file: how much of a scan should become text?
Convert the whole page a passage sits on, and the whole document when you expect to return to it. Converting less than a page saves nothing and loses context. Converting a full box of paper for the sake of one paragraph spends page allowance and checking time on text nobody will read.
A way to decide:
| Situation | Convert | Reason |
|---|---|---|
| One paragraph from a one-page letter | The page | The surrounding lines show what the paragraph refers to |
| One clause from a 40-page lease | The page with the clause and the pages it cross-refers to | A notice clause that points to "the address in Schedule 1" needs Schedule 1 |
| Several passages from one document over the life of a matter | The whole document, once | One conversion, one check, one text file to search afterwards |
| A single fact from a 300-page records bundle | The pages around the fact | Find it by reading or by page reference, then convert that range |
| Everything in a file being closed or transferred | Follow the firm's digitizing plan | A project with its own rules, not a paste job |
Cost in pages and in checking
Two budgets apply. The first is the tool's allowance. Schema Law's free plan, for example, takes one document a day of up to three pages and includes plain text among its exports, which covers a letter or a short notice. The pricing page shows what the paid plans allow. The second budget is your own time: text you have not compared with the page is text you cannot rely on, so every page converted is a page to check or to label as unchecked.
Splitting a page range out of a long scan
If you need pages 12 to 14 of a long PDF, extract them into their own file before uploading and keep the source page numbers in the file's name, such as Whitlock lease (scan) pp 12-14.pdf. The steps are in the post on free conversions, under cutting a scanned PDF down to a page cap.
Noting the source page on text that has left its scan
Text pasted into an email or a note carries no trace of where it came from. Add the source yourself at the moment you paste: the document's name and date, the page, and whether the text has been checked against the scan. Without that line, nobody can later confirm or correct what the passage says.
What a source note contains
Four items are enough.
- The document. Its description and date, as your file names it: "Calder Row Estates letter, 30 Sept 2026".
- The page. The page of the scan, plus the printed page number or Bates number if the document has one and they differ.
- The file. The name of the scanned PDF in the matter folder.
- The check. "Text from OCR, checked against scan" with initials and date, or "Text from OCR, not yet checked".
In a case note, that can be one bracketed line under the passage:
[Source: Calder Row Estates letter 30 Sept 2026, p. 1.
File: Whitlock - Calder Row rent notice 2026-09-30 (scan).pdf.
OCR text, checked against scan PS 2 Oct 2026.]In an email to a client, a shorter form reads more naturally: "Quoted from the landlord's letter of 30 September 2026, copy attached."
Why the check status belongs in the note
Do not count on OCR software to tell you which words it was unsure of. Schema Law's own homepage says recognition is strongest on clean print and weaker on faint faxes, stamps over text and handwriting, and that nothing marks a passage as uncertain. So the only record that a human compared the text with the page is the one you write. An unchecked passage is still useful as a pointer. It should not be forwarded as a quotation.
When the text is corrected after pasting
If you find an error after the text has already gone into a note, correct the note and say so: "Corrected 5 Oct 2026: date read 'Novernber', scan says 'November'." Then correct the .txt file too, so the next person who copies from it does not repeat the mistake. If the passage went out in an email, whether to send a correction is a judgment for the lawyer responsible.
Text that travels further
Passages get copied again: from the note into a chronology, from the chronology into a draft. Each copy sheds context. A source line that travels with the passage is the cheapest protection, and it lets whoever does the final read go straight to the right page of the right scan.