Book a Demo

The best handwriting recognition software in 2026: 8 tools compared

Handwriting recognition tools split four ways: document AI platforms, cloud OCR APIs, archive tools and note apps. Here's what 8 of them read, where they run and what they cost, and how to test one on your own pages.

Updated · 9 min read

Key takeaways

  • Handwriting recognition software reads handwritten text in scans, photos and PDFs and turns it into editable text or named fields. Business tools read the printed form around it too, and flag the values they're unsure about.
  • The tools fall into four groups: document AI platforms (Docsumo, Hyperscience, ABBYY Vantage), cloud OCR APIs (Azure Document Intelligence, Google Document AI, Amazon Textract), archive tools (Transkribus) and note apps (Pen to Print).
  • Handwriting covers far fewer languages than print: Azure Document Intelligence reads handwriting in 12 languages and Amazon Textract in English only.
  • Few vendors publish a handwriting accuracy figure, and those that do measure it their own way, so score field-level accuracy on your own worst pages.
  • Docsumo reads handwritten text alongside print and sends low-confidence fields to your own staff for review, with the source line highlighted.
On this page
  1. What handwriting recognition software does
  2. What makes handwriting recognition hard
  3. The 8 best handwriting recognition software tools
  4. Handwriting recognition software compared
  5. How to test handwriting recognition before you buy
  6. The bottom line
  7. Frequently asked questions

Handwriting recognition software, also called handwriting OCR, reads handwritten text in scans, photos and PDFs and turns it into editable text or structured data. For hand-filled business forms, a document AI platform such as Docsumo (our product) reads the handwriting and the printed form around it, then sends the values it's unsure about to a person. Developers build on a cloud API such as Azure Document Intelligence or Amazon Textract, archives use Transkribus, and for your own notes an app such as Pen to Print does the job.

Vendor details below come from each vendor's own website, checked in September 2026. Links are in the sources at the end.

What handwriting do you need to read?

For hand-filled forms in lending, insurance and AP files
Docsumo reads handwritten and printed text on the same page, checks the fields and sends only the values it's unsure about to your team.
For proofs of delivery in logistics
Docsumo reads handwritten arrival and departure times and tells a signed signature box from a blank one. Values it's unsure about still go to your team.
For large or sensitive volumes that can't leave your network
Docsumo runs in the cloud only. Hyperscience and ABBYY Vantage both run on-premises as well as in the cloud.
For an app you're building on Azure, AWS or Google Cloud
Docsumo if you'd rather not build review screens. Its API and webhooks return fields from handwritten and printed text, and your team checks uncertain values in Docsumo. If you build it yourself, Azure Document Intelligence reads handwriting in 12 languages and Amazon Textract in English only. Google Document AI reads it in a subset of its OCR languages.
For old letters, registers and archives
Docsumo is built for business documents. Transkribus is built for historical handwriting, with 300+ public models and training on your own pages.
For your own handwritten notes
Docsumo is built for teams that process business documents. Pen to Print turns notes into text, Word or Excel files on the web or your phone.

What handwriting recognition software does

Most handwriting in business files sits inside a printed form: a policy number written in boxes, a note in a comments field, a check mark in a box. Software reads each of those differently. Optical character recognition (OCR) reads the print, intelligent character recognition (ICR) the hand-printed capitals, handwriting recognition (HWR) the cursive, and optical mark recognition (OMR) the checkboxes. Business tools read print and handwriting in the same pass, then use the layout to work out which value belongs in which field.

Policy change request in four zones: printed header (OCR), boxed capitals (ICR), cursive note (HWR), checkboxes (OMR)
The boxes are printed in a dropout color that disappears on the scan, leaving ICR one capital per box to read.
  • Hand-filled forms
  • Scanned letters
  • Phone photos
  • Faxes
Handwriting recognition
  1. 01Clean up the image
  2. 02Find lines and fields
  3. 03Read print and handwriting
  4. 04Check and review
Text or fields in your system
From a handwritten page to usable data

The last step is where handwriting OCR tools differ most. Cloud APIs return text with a confidence score for your code to act on; document AI platforms also check each value against its field and send the doubtful ones to a person before the data moves on. For how the recognition itself works, see what handwriting recognition is and intelligent character recognition.

What makes handwriting recognition hard

Printed text looks the same on every page; handwriting never does. Five things cause most misreads.

  • Cursive and mixed styles

    Joined letters blur where one character ends and the next begins, and one form often mixes block capitals, cursive notes and a signature.

  • Old or damaged paper

    Faded ink, stains, creases and low-resolution scans add marks a model can mistake for strokes.

  • Boxes and ruled lines

    Characters touch or cross the printed box around them, and a rushed entry spills into the next field.

  • Numbers and shorthand

    A 1 read as a 7 or a misplaced decimal changes an amount, and shorthand such as PRN or mcg/kg means nothing without context.

  • Scripts and languages

    Handwriting support covers far fewer languages than print, and historical scripts need models of their own.

The 8 best handwriting recognition software tools

Docsumo is our product, so we've put it first and said plainly what it doesn't do. The others are grouped by what they're built for.

Document AI platforms

1. Docsumo

Document AI platformOur product
Docsumo is an intelligent document processing (IDP) platform for lending, insurance and accounts payable teams. It reads handwritten and printed text on the same page, such as a hand-filled ACORD form or the arrival and departure times written on a proof of delivery, and turns them into named fields. It isn't a desktop tool for editing PDFs or making searchable files, and it runs in the cloud only.
Reads
Printed and handwritten text on any business document; covers 250+ document types
Review
Fields below the confidence threshold you set for each field go to your own reviewers, with the source line highlighted, and their corrections improve the model
Output
Named fields and tables through an API and webhooks, or an Excel export
Pricing
A free 14-day trial for up to 1,000 pages; Business and Enterprise plans are quoted
Best fitLending, insurance and AP teams whose files mix print and handwriting
  • 95%+of documents processed straight through, without manual review
  • <5 minper document, down from 2+ hours
  • 3,000+hours a month saved at Arbor

2. Hyperscience

Document AI platform
Hypercell, Hyperscience's platform for back-office documents in large enterprises and government, pairs its own models with targeted human review: people see only the fields or characters that fall below the accuracy target you set. Hyperscience says its models reach 98% accuracy on handwriting and cursive, by its own measure.
Reads
Structured forms, invoices, contracts and long documents, including handwriting and cursive
Runs on
SaaS on AWS or Google Cloud, on-premises or air-gapped, with FedRAMP High authorization for government
Pricing
Not published; volume-based rather than per user
Best fitLarge enterprises and government agencies with high volumes or strict hosting rules. See Docsumo vs Hyperscience.

3. ABBYY Vantage

Document AI platform
ABBYY's intelligent document processing platform, built on pre-trained extraction models it calls Skills and set up with low-code tools. ABBYY says Vantage processes handwriting, barcodes and checkboxes, with human-in-the-loop review. It still sells its older capture platform, FlexiCapture, which reads handwritten data with ICR.
Reads
Structured, semi-structured and unstructured documents, including handwriting, barcodes and checkboxes
Runs on
ABBYY Cloud in Europe, the US or Australia, on-premises, or a private cloud on Microsoft Azure
Pricing
Not published
Best fitEnterprises that want ABBYY's cloud or on-premises deployment and have IT teams to run it. See Docsumo vs ABBYY FlexiCapture.

Cloud OCR APIs

These return text, fields and confidence scores as JSON, priced per page. Review screens, checks and delivery to other systems are yours to build.

4. Azure Document Intelligence

Cloud OCR API
Microsoft's OCR and document extraction service, now called Azure Document Intelligence in Foundry Tools (formerly Form Recognizer). Its read and layout models return printed and handwritten text with its position on the page, and prebuilt and custom models pull out fields.
Handwriting
12 languages in API v4.0, including English, Spanish, French, German, Chinese, Japanese and Arabic; printed text in 300+
Output
JSON through a REST API and SDKs, with Document Intelligence Studio for testing
Pricing
Read $1.50 per 1,000 pages; a free tier of 500 pages a month (published)
Best fitDevelopers on Azure, including teams that need to run it in containers. See Docsumo vs Azure Document Intelligence.

5. Google Document AI

Cloud OCR API
Google Cloud's document service. Its Enterprise Document OCR processor reads printed and handwritten text, and a font-style add-on flags which words are handwritten. Google has deprecated Document AI's human-in-the-loop review feature, so plan your own review step.
Handwriting
A subset of the 200+ languages its OCR reads; Google's language table lists which
Output
JSON with text, layout, tables and entities
Pricing
Enterprise Document OCR $1.50 per 1,000 pages; add-ons $6 per 1,000 pages (published)
Best fitDevelopers on Google Cloud with documents in many languages. See Docsumo vs Google Document AI.

6. Amazon Textract

Cloud OCR API
AWS's machine learning service for text, handwriting, forms, tables and signatures, with specialist APIs for expenses, IDs and mortgage packages. Its human review add-on, Amazon Augmented AI, is no longer open to new customers, so plan your own review step.
Handwriting
English only; printed text, forms and tables in English, German, French, Spanish, Italian and Portuguese
Output
JSON blocks with positions and a confidence score for each item
Pricing
Text $1.50 per 1,000 pages, forms $50 per 1,000 pages; a 3-month free tier for new AWS customers (published, US West)
Best fitDevelopers on AWS reading English handwriting on forms. See Docsumo vs Amazon Textract.

Archive and research tools

7. Transkribus

Archive tool
A platform for historical handwriting from READ-COOP, a European cooperative, used by archives, libraries, researchers and genealogists. It reads handwriting from any century, including old scripts such as Kurrent, and can train a model on your own documents from as few as 50 transcribed pages.
Reads
Handwritten and printed text in 100+ languages, with 300+ public models trained by its community
Output
TXT, DOCX, PDF, TEI-XML, PAGE XML or ALTO, plus a REST API; data is processed and stored on its own servers in Austria
Pricing
A free plan with 50 credits a month, at 1 credit per handwritten page; Scholar €99 a year, Team €449 a year (published)
Best fitArchives, libraries and researchers transcribing historical documents

Note apps

8. Pen to Print

Note app
A handwriting-to-text app for notes, scans and PDFs, on the web and on iOS and Android. It turns handwritten pages into editable files, and its site claims 98.7% word accuracy.
Reads
Handwritten notes in print or cursive, scans and PDFs; tables, math and forms on its All-inclusive plan
Output
Text, Word, Excel, LaTeX and searchable PDF; an API
Pricing
10 free pages, then from $9.99 a month for 100 pages, or $59.99 a year (published)
Best fitStudents, teachers and small offices digitizing notes

Handwriting recognition software compared

ToolReadsRuns onReview queuePricing
Document AI platforms3 tools
DocsumoOur productFormsPrintHandwritingTablesCloudBuilt in, thresholds per fieldFree trial; quoted plans
HyperscienceFormsHandwritingCursiveLong documentsCloud, on-premises, air-gappedBuilt inNot published
ABBYY VantageFormsHandwritingBarcodesCheckboxesCloud, on-premises, private cloudBuilt inNot published
Cloud OCR APIs3 tools
Azure Document IntelligencePrintHandwriting in 12 languagesTablesAzure, containersNot described$1.50 per 1,000 pages (read)
Google Document AIPrintHandwritingCheckboxesMathGoogle CloudDeprecated$1.50 per 1,000 pages (OCR)
Amazon TextractPrintEnglish handwritingFormsTablesAWSClosed to new customers$1.50 per 1,000 pages (text)
Archive tools1 tool
TranskribusHistorical handwritingPrint100+ languagesCloud (Austria)Editor for correctionsFree 50 credits/mo; from €99/yr
Note apps1 tool
Pen to PrintNotesCursiveTablesMathWeb, iOS, AndroidNot describedFrom $9.99/mo

How to test handwriting recognition before you buy

Few vendors publish a handwriting accuracy figure. Hyperscience and Pen to Print do, each measured their own way; Docsumo, Microsoft, Google and AWS don't. The number that counts is the one you measure on your own pages.

  1. Build a test set from your worst pagesPull a few hundred real documents, including the oldest paper, the most hurried writing and the most crowded forms, in the mix you actually receive.
  2. Pick the metric and the target firstScore field-level accuracy, the share of fields that come out exactly right, not character error rate: one wrong digit makes a whole amount wrong. Set a target for each field before you see any results. More in how OCR accuracy is measured.
  3. Score every tool against the same answersHave a person key the correct value of every field first, then run each tool on the same pages and compare field by field.
  4. Price the review, not just the softwareCount the fields each tool sends to review at your threshold, multiply by the time to fix one and your reviewers' hourly cost, and add the software price.
Six invoice fields scored against per-field thresholds: five auto-accepted, a handwritten PO number at 0.71 sent to review
Values that clear their field's threshold go straight through; the handwritten PO number falls short, so a reviewer checks it against the highlighted spot on the page.

The bottom line

Pick the kind of tool before the vendor. For hand-filled forms in business files, use a document AI platform: Docsumo fits lending, insurance and accounts payable teams that want handwriting read, checked and sent on, with only the unsure fields left for their own staff. Hyperscience and ABBYY Vantage suit enterprises that need on-premises deployment. Developers can start with Azure, AWS or Google, archives with Transkribus, and note-takers with Pen to Print. For OCR tools beyond handwriting, see the best OCR software.

Book a demo and bring a few of your hardest handwritten pages, or start a free trial.

Frequently asked questions

What is the best handwriting recognition software?

It depends on what you're reading. For hand-filled business forms that need checking, use a document AI platform such as Docsumo, Hyperscience or ABBYY Vantage; to build handwriting recognition into an app, Azure Document Intelligence, Amazon Textract or Google Document AI; for historical documents, Transkribus; and for your own notes, Pen to Print. Test any of them on your hardest pages.

Is there free handwriting recognition software?

Several tools are free up to a limit. Transkribus includes 50 credits a month (about 50 handwritten pages), Azure Document Intelligence has a free tier of 500 pages a month, Pen to Print gives you 10 free pages, and Docsumo has a free 14-day trial for up to 1,000 pages. Tesseract is free and open source, but its own FAQ says it won't work well on handwriting because it's designed for printed text.

What software can read old handwriting?

Transkribus is built for it. It reads handwriting from any century, including old scripts such as Kurrent, offers 300+ public models trained by its community, and can train a model on your own documents from as few as 50 transcribed pages.

Can OCR read cursive handwriting?

Yes. Modern handwriting recognition reads cursive as well as print, but cursive is harder because joined letters blur where one character ends and the next begins, and accuracy drops further on faded or low-resolution scans. Business tools send low-confidence values to a person; see human-in-the-loop review.

Which languages can handwriting recognition read?

It depends on the tool, and handwriting lists are much shorter than print lists. Azure Document Intelligence reads handwriting in 12 languages (API v4.0), Amazon Textract in English only, and Google Document AI in a subset of the 200+ languages its OCR reads. Transkribus lists 100+ languages, including historical scripts.

Is handwriting recognition the same as handwriting identification?

No. Handwriting recognition reads what was written and turns it into text. Handwriting identification is forensic work that compares samples to decide who wrote them, and none of the tools in this guide describe doing it. Some can detect handwriting, though: Google Document AI's font-style add-on flags which words are handwritten.

Sources

  1. Microsoft: Azure Document Intelligence overview
  2. Microsoft: Read and layout language support (handwritten text)
  3. Microsoft: Azure Document Intelligence pricing
  4. Google Cloud: Enterprise Document OCR
  5. Google Cloud: Document AI processor list (supported languages)
  6. Google Cloud: Document AI pricing
  7. Google Cloud: Document AI deprecations (human-in-the-loop)
  8. AWS: Amazon Textract features
  9. AWS: Amazon Textract FAQs
  10. AWS: Amazon Textract pricing
  11. AWS: Use Amazon Augmented AI with Amazon Textract (not open to new customers)
  12. Hyperscience: Hypercell platform
  13. ABBYY: Vantage
  14. ABBYY: FlexiCapture
  15. Transkribus: home page
  16. Transkribus: AI text recognition
  17. Transkribus: plans
  18. Pen to Print: home page
  19. Pen to Print: pricing
  20. Tesseract OCR: FAQ

First published . Last updated .

More on OCR & IDP →

See Docsumo read your own documents

Bring a few real samples. We'll show the fields extracted, the checks that ran and what a reviewer would see.

Book a demo →or get a free trial (1,000 pages)