Other programs compared
There is a whole range of programs that read Fraktur or help with correcting. None is the best at everything; most of them can be combined with Fraktur-Korrektor. This page is meant to make the choice easier – it is not advertising, and it is a snapshot: all statements come from the programs' official pages, as of 9 October 2026, and may change. We have not tried the other programs on books ourselves: Fraktur quality is described, not measured – serious figures only exist per book and per scan. What only the maker says is marked according to ….
Which tool for what
- You have a scanned book and want to read and correct it: read it in with the built-in Tesseract first. If the traffic light is green, you are done. Otherwise Transkribus – convenient, good at Fraktur, but online and, beyond a certain volume, paid.
- You own ABBYY FineReader, or you have a PDF that was already recognised with it or by a library: Fraktur-Korrektor takes over the text layer of a searchable PDF in seconds (using the existing text).
- You work at an institution that runs eScriptorium or OCR-D, or you want to train your own models: those are the right tools for that – with a server installation and a learning period. Their result (PAGE XML, ALTO, hOCR) can be read on here (text from a library).
- You want to improve the scan, not the text: Fraktur-Korrektor splits double pages, straightens and removes dark margins; for curvature and stains there is ScanTailor.
- You already have text (EPUB, text file, Transkribus export) and want to read it against the scan: that is exactly what Fraktur-Korrektor is for (Opening a book).
What fits together: Transkribus → Page XML → read in here. ABBYY or OCRmyPDF → searchable PDF → use the existing text here. Tesseract, gImageReader, OCR-D, eScriptorium → hOCR or ALTO → Take in recognised text here. An EPUB (for instance an earlier run of your own) → put it next to the PDF → open here. Fraktur-Korrektor itself writes text files per page, a PDF with a text layer that any other program can read, and an e-book.
The table
How to read it: Fraktur means whether and how the program recognises Fraktur print; Correcting means how comfortably you fix errors afterwards; Offline means whether your scans stay on your computer.
| Program | Price | Fraktur | Correcting | Offline / privacy | Learning curve | Export | Platform |
|---|---|---|---|---|---|---|---|
| ABBYY FineReader PDF (version 16) | Subscription only. Windows: Standard €99/year, €267 for 3 years or €16/month; Corporate €165/year, €446 for 3 years or €24/month. Mac: €69/year or €186 for 3 years | Language “Old German” for Textur, Fraktur and Schwabacher according to the version 16 help. ABBYY advises selecting only this one language | Verification dialog jumps from uncertain word to uncertain word, with an enlarged image detail; user dictionary (according to the manual) | Yes – runs on your own computer; internet only for activation | Low – an office program | Searchable PDF (also PDF/A), Word, Excel, text, HTML, EPUB and more | Windows 10/11, Mac (Fraktur on the Mac too, according to an ABBYY help article) |
| Transkribus (READ-COOP, Innsbruck) | Free: 50 credits/month; Scholar €99/year with 900 credits; 250 credits on their own €59.50; a printed page 0.5 credit, handwriting 1 credit | Very good: public models for German Fraktur with 0.5–2.2 % character errors on their own test pages (figures from the model pages) | Web editor, line in the image; for a whole book more laborious than a keyboard program | No – images are processed and stored on servers in Austria. A version for institutions' own servers is in early access (price on request) | Medium – account, collection, jobs, export | Page XML, text, Word, PDF on all plans; ALTO, METS, TEI, Excel paid only | Browser |
| Tesseract (5.5.3, July 2026) | Free, Apache licence | Usable to good with clean, straight scans and a Fraktur model (frak2021 from Mannheim University Library, deu_latf); sensitive to skew, curvature, phone photos – our experience, see traffic light |
None – recognition only | Yes | High (command line); built into Fraktur-Korrektor: one click | Text, hOCR, ALTO, PAGE XML, PDF with text layer | Windows, Mac, Linux |
| OCR4all with LAREX (0.6.1, January 2022) | Free, MIT licence | Built for early prints; recognition with Calamari; own models can be trained | Built-in correction step for training text, with an on-screen keyboard for special characters | Yes – runs in Docker on your own computer | High – Docker, multi-step workflow, meant for projects | PAGE XML, text | Docker (Windows, Mac, Linux) – no longer developed, maintenance only; work continues in its component LAREX (according to the README) |
| eScriptorium with Kraken (26.07, September 2026) | Free, MIT (eScriptorium) and Apache (Kraken) | Models for German prints according to the guide by Mannheim University Library (german_print); own models can be trained |
Web editor, line in the image, on-screen keyboard | Yes, if you or your institution run it yourselves (server) | High – server installation; as a user of an instance, medium | Text, PAGE XML, ALTO; TEI if the instance has switched it on | Server on Linux, on Windows via WSL (according to Mannheim University Library); Kraken alone: Linux, Mac |
| OCR-D (3.13.3, September 2026) | Free, Apache licence | Toolbox for libraries, meant for German prints of the 16th to 19th century (DFG-funded); quality depends on the chosen workflow and model | No interface of its own | Yes | Very high – command line, Docker, Ubuntu | PAGE XML, ALTO, METS | Linux/Docker; Windows only via WSL, Mac via Homebrew (according to the setup page) |
| gImageReader (3.4.3, August 2025) | Free, GPL | The same recognition as Tesseract; you install the Fraktur model yourself | Text next to the image, post-processing with spell checking | Yes | Low to medium | Text, hOCR, PDF (from hOCR) | Windows, Linux (Mac not mentioned in the README) |
| PoCoTo (CIS Munich) | Free, BSD licence | No recognition – post-correction of ABBYY or Tesseract results only (ABBYY XML, hOCR) | Batch correction of whole error series (research tool from the IMPACT project) | Yes (Java); the error profile for correction suggestions comes either from a web service or, for experts, from your own computer | Medium to high | Text (whole or per page), back into the files read in | Java (Windows, Mac, Linux) – last change May 2017, effectively unmaintained |
| Fraktur-Korrektor (this program) | Free, GPL | Recognises nothing itself: Tesseract built in, Transkribus via import; a traffic light rates the quality | That is its purpose: keyboard, red words, batch correction, dictionary per era | Yes | Low | Text per page, PDF with text layer, e-book (EPUB) | Windows, Mac, Linux |
Sources
All retrieved on 9 October 2026.
- ABBYY: prices https://pdf.abbyy.com/pricing/; system requirements and formats https://pdf.abbyy.com/specifications/; Gothic/Fraktur in version 16 https://help.abbyy.com/en-us/finereader/16/user_guide/gothicrecognition/; verification dialog (version 15 manual) https://help.abbyy.com/en-us/finereader/15/user_guide/checkingtext/; Mac article https://support.abbyy.com/hc/en-us/articles/4417503390611 (the page refused automated retrieval – only title and search snippet checked).
- Transkribus: plans and credits https://www.transkribus.org/plans, https://help.transkribus.org/credit-system; data storage https://help.transkribus.org/is-my-data-safe, own servers https://www.transkribus.org/on-prem; models https://www.transkribus.org/models/transkribus-print-multi-language-dutch-german-english-finnish-french-swedish-etc, https://www.transkribus.org/models/german-fraktur-19th-20th-century, https://www.transkribus.org/models/german-fraktur-18th-20th-century; export https://help.transkribus.org/downloading.
- Tesseract: https://github.com/tesseract-ocr/tesseract/releases; Windows installer and frak2021 https://github.com/UB-Mannheim/tesseract/wiki.
- OCR4all: https://www.ocr4all.org/, https://github.com/OCR4all/OCR4all (maintenance-mode notice), https://github.com/OCR4all/OCR4all/releases, https://www.ocr4all.org/guide/user-guide/workflow.
- eScriptorium and Kraken: https://gitlab.com/scripta/escriptorium (versions), export https://escriptorium.readthedocs.io/en/latest/export/, Kraken https://kraken.re/main/getting_started.html, guide by Mannheim University Library https://ub-mannheim.github.io/eScriptorium_Dokumentation/.
- OCR-D: https://ocr-d.de/en/, https://ocr-d.de/en/about, https://ocr-d.de/en/setup, https://github.com/OCR-D/core.
- gImageReader: https://github.com/manisandro/gImageReader.
- PoCoTo: https://github.com/cisocrgroup/PoCoTo (licence, last change in the commit history), manual https://github.com/cisocrgroup/Resources/blob/master/manuals/pocoto-manual.md, download https://www.cis.lmu.de/ocrworkshop/data/pocoto/.