Is There a Scanner That Converts Handwriting to Text?
By Schema LawPublished Updated
No scanner converts handwriting to text by itself. A scanner photographs the page, and recognition software, either bundled with the scanner or run on the file afterwards, turns the picture into characters. So the question to ask when buying is which software will read the handwriting, and how well it reads yours.
Is there a scanner that converts handwriting to text?
There is no scanner that converts handwriting to text as a hardware function. Every scanner does one job: it records the page as an image. Text appears only when recognition software reads that image. Some scanners run such software for you at scan time, which makes the box look as if it did the reading.
That split has a useful consequence. You can separate the two decisions. The scanner you already own may make a perfectly good image of a handwritten page, and the reading can be done by a different program than the one that came in the box. The post on what an OCR text scanner does explains where the recognition runs when a scanner or copier offers a searchable PDF button.
You still read names, dates, amounts, and signatures yourself, on a legal pad and on a letter a client handed you. Keep the image. It is your documentary evidence. Note the machine and the day, and you have the start of a chain of custody. Use the scan for authentication of faint ink, and judge relevance on that same image.
For a handwritten page the path has three stages:
- Capture. The scanner makes an image file: a PDF, a JPEG or a TIFF.
- Recognition. Software looks at the image and proposes characters and words.
- Checking. A person reads the proposed text against the page.
Hardware owns stage one only. If the text that comes out is wrong, the cause is either a poor image, which the scanner can fix, or a reader that struggles with the hand, which no scanner setting will fix.
To convert a scanned handwritten image you already have, you need no hardware at all. Open the file in recognition software or upload it to a service, export the text, and proofread it. The post on converting handwriting to text gives the procedure.
What a scanner's bundled software does with handwritten pages
Bundled scanner software is built around printed documents, and at least one maker says outright that handwriting is outside its comfort zone. The software will still run on a handwritten page and still produce text. What it produces has to be judged on your own pages, because the manual makes no promise.
Ricoh's ScanSnap documentation is a clear example. Its help page on the OCR function for converting scans to Word, Excel and PowerPoint files lists "documents including handwritten characters" first among the types of documents and characters that "may not be recognized correctly". The same list includes characters smaller than 10 point, skewed documents and documents written in a language other than the one specified. The page adds that such documents "may be recognized if you scan them by changing the color mode or improving the image quality".
Read your own scanner's manual for the equivalent list before you plan a project around its software. Look for three things:
- A statement about handwriting. Either a caution like the one quoted, or a named handwriting feature with its conditions.
- The language setting. Recognition software reads against the language it is told to expect.
- What the output is. A searchable PDF hides the recognized text behind the image, so errors are invisible until you search or copy. A Word or text export puts the errors where you can see them.
That third point matters more for handwriting than for print. A searchable PDF of a handwritten page can look finished and hold a text layer that is mostly wrong. Copy a paragraph out and read it before you trust a search of that file.
Smart pens and reusable notebooks: devices that capture handwriting as you write
Smart pens and reusable notebooks are the other hardware sold for handwriting. A smart pen records the strokes while you write on paper. A reusable notebook is ordinary writing on a special page that you photograph with a companion app. Neither is a scanner, and both still rely on software to produce text.
The two differ in what they hand the software:
| Device | What it records | What reads it |
|---|---|---|
| Smart pen | The pen's movement, stroke by stroke, as you write | The pen maker's app |
| Reusable notebook | A phone photo of the finished page | The notebook maker's app or a service it sends to |
| Sheet-fed or flatbed scanner | An image of the finished page | Bundled or separate recognition software |
A stroke record carries the order and direction of each line, which a picture does not. The post on handwritten text recognition explains why that helps a reader.
Both devices only help with handwriting you have not written yet. They do nothing for the legal pads in a closed file or a letter a client wrote. Both also tie the notes to the maker's app, so ask the same questions you would ask of any app: where the pages are stored, what the transcription sends over the network, and how you export. For one reusable notebook in detail, see the post on Rocketbook handwriting to text, which is built from the company's own pages.
Does a more expensive scanner read handwriting better?
A more expensive scanner does not read handwriting better, because the scanner does not read. Money buys speed, a larger feeder, two-sided scanning and sturdier paper handling. Those matter for volume. Past the point where the image is clean and sharp, recognition of handwriting depends on the software and on the hand.
Image quality still sets a floor. Faint pencil scanned in pure black and white can vanish before any software sees it, and a cheap scanner at sensible settings will beat an expensive one at bad settings. The capture choices that matter for handwritten pages, including color mode, ink and paper, are covered in the post on scanning handwriting for recognition.
A fair way to find out whether new hardware would help:
- Pick five handwritten pages from real files: neat print, joined writing, pencil, a ruled yellow pad and a page with margin notes.
- Scan them on the scanner you have, in grayscale or color at 300 dpi.
- Run the files through the recognition software you are considering.
- Count the wrong words on each page.
If the images are sharp and the text is still poor, a new scanner will not change the result. Different software might, and the post on testing a handwriting converter on your own pages shows how to compare.
Choosing between a scanner, a tablet and a phone for handwritten pages
Choose by where the handwriting is now. Paper that already exists needs a scanner or a phone camera. Notes you have yet to take can be written on a tablet and never touch paper. Volume and page condition decide between the scanner and the phone.
| Situation | Device | Why |
|---|---|---|
| A box of legal pads from closed files | Sheet-fed scanner | The feeder takes a stack; pages come out as one ordered file |
| A bound notebook or fragile original | Flatbed scanner | The page lies flat and nothing is pulled through rollers |
| One page handed to you at a client's kitchen table | Phone | It is the device you have with you |
| Notes at next month's deposition preparation | Tablet with a stylus | The app records strokes and can convert them |
| A mix of all four | Scanner for the backlog, tablet going forward | Each device does the part it suits |
Tablets and their apps are a subject of their own; the post on tablet handwriting to text apps covers them as a class.
Once you have an image, the device no longer matters. A PDF, PNG or JPEG of a handwritten page can go to any recognition tool. Schema Law reads uploaded pages into text and exports them in the formats listed on the homepage; its own site says recognition is weaker on handwriting and that nothing marks a passage as uncertain. Whatever reads the page, plan time for a person to compare names, dates and figures with the original.