Convert a typewritten document to text

[#1]

I am looking for some software that can convert an old document written on a typewriter to a text file you can edit.  Typed documents have some unique features:

• There is only one font
• If a character is slightly wrong or misaligned then that same character is always the same slightly wrong or misaligned.

So you might think this would be a easy task for OCR (Optical Character Recognition) software and you would get even better results if the software is prepared to let you teach it how to recognise individual letters.  But no.  You end up with something that is ridiculously over-complicated with copious errors in a whole range of different fonts and with peppered with other features that could not be achieved using a typewriter.  And I have not found any OCR software that is prepared to learn.

Any suggestions?

[#2]

I think all you can do is try different OCR software.

Not used it for years, someone may have suggestions on a up to date good application to use.

Depends on the size of the document, but it is never 100% in my experience so you will have to go through it correcting errors manually.

I suppose using good software means there are less of them to deal with.

[#3]

You should be able to correct font variances by simply opening the document in a word processor, highlighting the complete document and selecting your preferred font.

With respect to OCR accuracy, have you tried expanding the typed text by scanning a portrait page and printing it again in landscape? it will need two pages to get the complete portrait page but it might give the OCR a bit more detail to go at.

Failing that you may just have to suffer the pain using a word processor or seek professional help, I guess it depends on the size of the document when deciding if that's practicable.

[#4]

How big ( number of pages ? ) is the document?  Is it "plain typing"   as in a manuscript/novel or is it tabulated etc?

You may find it cheaper to employ some self employed office typewriting person, than to squander oondles of money on various OCR softwares that you may never need to use again..

[#5]

---

And I have not found any OCR software that is prepared to learn.

Any suggestions?

---

Have you tried Adobe Acrobat? Not sure about learning but Omnipage can be taught. Both have time limited trial versions.

[#6]

So far I have tried Iris OCR and, since my original post, ABBYY FineReader.  The ABBY software produced a much better result.  You can tell it the input is typewritten and it then produces output that is all in one font and with a basic and simple layout.  But it has no learning capability that I have found so far.  I'll look into the other suggestions.   

[#7]

OmniPage Pro is definitely the one to use. See http://supportcontent.nuance.com/omnipage/18/doc/OP18Guide.pdf

[#8]

I had a go with OmniPage but was disappointed in the result.  Some of the resulting text was laid out in document format but random sections were placed in text boxes.  It's another example of software trying to be too clever and thereby giving a stupid result.  Typed and printed documents typically have a very simple layout and the last thing you want is some OCR software that renders this more complicated then the original.  If there is a "Don't put text in boxes" setting or a "Don't create random lines" option then I failed to find them.  

[#9]

If you go to Options then you can set the layout to be the simplest possible

[#10]

I was trying to deal with a document that had a double-line margin around, and quite close to, the text.  I wanted to ignore this but the Nuance software got fixated with it, seeing it either as characters or rendering it as bits of vertical line - which appeared even in a .rtf format output.  I would have got a better result if I had printed out the pages, cut-off these margins with a pair of scissors and re-scanned them.  So for me, Omnipage is in the "stupid result by trying to be too clever" category.    

[#11]

Acrobat has a cropping feature and a redaction feature so you could scan your document using Acrobat, crop out the lines or redact them then save the result as a PDF. Omnipage can open a PDF and do the OCR.

[#12]

Yes I see what @daveplus means.

A basic cropping feature will do what you want, many programs have them.

More hassle than if you didn't need to, but less if you tried your manual way. 

[#13]

Why do you need this document as text, anyway?

A print of an image is just as good as a typewritten document.

[#14]

@VileReynard  It seems you have jumped in with both feet.... the first line of the original post gives a clue...

I am looking for some software that can convert an old document written on a typewriter to a text file you can edit.

G

M

T

Detect language Afrikaans Albanian Amharic Arabic Armenian Azerbaijani Basque Belarusian Bengali Bosnian Bulgarian Catalan Cebuano Chichewa Chinese (Simplified) Chinese (Traditional) Corsican Croatian Czech Danish Dutch English Esperanto Estonian Filipino Finnish French Frisian Galician Georgian German Greek Gujarati Haitian Creole Hausa Hawaiian Hebrew Hindi Hmong Hungarian Icelandic Igbo Indonesian Irish Italian Japanese Javanese Kannada Kazakh Khmer Korean Kurdish Kyrgyz Lao Latin Latvian Lithuanian Luxembourgish Macedonian Malagasy Malay Malayalam Maltese Maori Marathi Mongolian Myanmar (Burmese) Nepali Norwegian Pashto Persian Polish Portuguese Punjabi Romanian Russian Samoan Scots Gaelic Serbian Sesotho Shona Sindhi Sinhala Slovak Slovenian Somali Spanish Sundanese Swahili Swedish Tajik Tamil Telugu Thai Turkish Ukrainian Urdu Uzbek Vietnamese Welsh Xhosa Yiddish Yoruba Zulu

Afrikaans Albanian Amharic Arabic Armenian Azerbaijani Basque Belarusian Bengali Bosnian Bulgarian Catalan Cebuano Chichewa Chinese (Simplified) Chinese (Traditional) Corsican Croatian Czech Danish Dutch English Esperanto Estonian Filipino Finnish French Frisian Galician Georgian German Greek Gujarati Haitian Creole Hausa Hawaiian Hebrew Hindi Hmong Hungarian Icelandic Igbo Indonesian Irish Italian Japanese Javanese Kannada Kazakh Khmer Korean Kurdish Kyrgyz Lao Latin Latvian Lithuanian Luxembourgish Macedonian Malagasy Malay Malayalam Maltese Maori Marathi Mongolian Myanmar (Burmese) Nepali Norwegian Pashto Persian Polish Portuguese Punjabi Romanian Russian Samoan Scots Gaelic Serbian Sesotho Shona Sindhi Sinhala Slovak Slovenian Somali Spanish Sundanese Swahili Swedish Tajik Tamil Telugu Thai Turkish Ukrainian Urdu Uzbek Vietnamese Welsh Xhosa Yiddish Yoruba Zulu

Text-to-speech function is limited to 200 characters

Options : History : Feedback : Donate

Close

[#15]

Fair enough...

How about pre-processing the image scans?

e.g. Increase the contrast so you have just black & white - no greys; then remove speckles (dirt etc) before asking OCR to do its thing?