Translating Instruction Manuals into English

[Adam HarrisParticipant @adamharris13683]

I have been investigating the easiest way to convert a SCANNED image of a pdf document in German with imbedded diagrams  (some containing their own text) into an English document retaining the original page formatting. There seems to be no software out there and no AI app that can handle this, and the best I can come up with is using Adobe to convert the scanned pdf image into an editable pdf (sort of pdf) page by page and then do a lot of manual cut and pasting back and forth into google translate. Very tedious when the manual is 140+ pages. I am wondering if anyone here has found a better method. Incidentally using Chat GPT etc for this is an exasperating venture down rabbit holes of mistruths and can/can’t, but that is a separate topic…

[Adam HarrisParticipant @adamharris13683]

Needless to say I have checked out the main software programs for translation but I have not found any that can do this task.

[peak4Participant @peak4]

Obviously I’ve not seen your document, but this is from Google Translate, using the Image option, of a random page in German, but would still need doing 140 times.
There are free web interfaces where you can upload individual pdfs, and then download a combined single pdf file.

I’m presuming you tried pdfguru.com; I remembered I had a German copy of a Myford ML10 manual; It claimed to have OCR translated that OK, but wanted to charge me for the download. (99p)

Bill

[IanTParticipant @iant]

I’m afraid I don’t have a solution to your specific OCR problem Adam.

However, for simple jobs I have used MS OneNote to read German text from photos – you simply right click on the photo and click “Copy Text from Photo”

I then paste the text into Google Translate and copy the translation into Word to format it.

I photograph any pages I want with my phone. Not ideal (or quick) but it certainly works for smaller jobs and you don’t need a scanner.

Regards,

 

IanT

[Dick HParticipant @dickh]

Depends what the manual is, sometimes you get lucky and there is an English version out there. Else Google translate.

[Adam HarrisParticipant @adamharris13683]

Bill thanks I have not tried the image option in google translate, looks like it handles embedded diagrams+text perfectly. Will check it out and revert. I did try pdfguru but it was unable to deal with the document being a scanned image rather than a pdf, or at least not automatically and so no better than Adobe itself

[Adam HarrisParticipant @adamharris13683]

Ian thanks, I will take a look at MS OneNote, maybe it will be less cutting and pasting than Adobe

[larry phelan 1Participant @larryphelan1]

Try translating China type manuals into English!

Interesting, to say the least !, but not always helpful.

[David JuppParticipant @davidjupp51506]

Based a recommendation from a colleague in Germany, I use DeepL for my translation needs – it copes well with technical stuff, but be warned some things do not translate literally.  You’ll need to overview the output to check that it both makes sense and that it reads well (that is true for any translations tool).

[larry phelan 1]

On larry phelan 1 Said:

Try translating China type manuals into English!

Interesting, to say the least !, but not always helpful.

I don’t really follow you there, Larry.

Do you mean translate using a human brain or translate using translation software?

[Adam HarrisParticipant @adamharris13683]

Bill, your advice is brilliant. Scan pages and save as a collection of jpeg images, then use Google Image translator, then print translated result to Microsoft Print As PDF, and do final editing touches if needed on that pdf. Big mistake was to scan and save as a pdf file and then try translating the pdf. Many thanks, Adam

[Clive FosterParticipant @clivefoster55965]

Allegedly the later versions of ReadIRIS can do that sort of multi language OCR/Translation and keep all the diagrams in the right place.

I have an older version I’m not sufficiently impressed with to learn how to use properly. Retaining formatting for diagrams was iffy in my hands.

Clive

[Adam Harris]

On Adam Harris Said:

Bill, your advice is brilliant. Scan pages and save as a collection of jpeg images, then use Google Image translator, then print translated result to Microsoft Print As PDF, and do final editing touches if needed on that pdf. Big mistake was to scan and save as a pdf file and then try translating the pdf. Many thanks, Adam

Glad it worked OK, and don’t forget that there are free on-line sites where you can upload a series of pdf pages and then download  them as a single document.
I used to use Acrobat, but that was a Windows XP version.
I appreciate that there is paid for software to do it, but I’ve not managed to find a good stand alone free package.

For basic annotation Foxit Reader (free) works OK on pdf files; I use it for filling in forms, to save printing, writing and then scanning.
Also for free jpg manipulation and annotation, have a look at Windows only – Faststone Image Viewer (Free/Shareware, I chuck a few quid in the pot when a significant new version is released)
It’s no Photoshop, but you can clean up, change contrast etc as well as annotating.

Bill

[Adam HarrisParticipant @adamharris13683]

Thanks for that further advice Bill. I have not figured out how to load images in a group into Google Images Translation rather than one by one  – maybe it only likes one at a time. I am using Windows 10

[Adam Harris]

On Adam Harris Said:

… Big mistake was to scan and save as a pdf file and then try translating the pdf. Many thanks, Adam

Good result!  Converting documents is rather like many machining jobs in that the order operations are applied makes a difference. PDF is a particularly nasty start point.

docx, rft, and other word processor files have a more-or-less well defined structure that’s relatively simple to unpick.  Text can be extracted without much bother.  PDF isn’t like that!  It’s a container format, only designed to be printable.   The contents aren’t organised to be edited or extractable.   Though not completely impossible, PDFs are difficult to edit and analyse.  The various tools available have mixed results, ranging between “worked perfectly” to “total failure” depending on how the PDF was generated in the first place.

I’m not surprised CHAT-GPT got into a tangle trying to explain what might be done – there are a multitude of options and many of them conflict.  It’s the asking for directions in darkest Zummerzet joke – the long complicated answer ends up with  “don’t start from here zurr!”

Avoid working with PDFs if at all possible!   Bill’s advice is excellent – it provides jpgs that OCR can chew on, ready for language translation later.

Dave

 

 

 

 

[larry phelan 1Participant @larryphelan1]

No Bill, I mean just trying to make sense of the “instructions” supplied with many of the machines and gear coming from China. The items work OK, but reading through the “guides” is an entertainment in itself.

The  “Instructions” which came with my rotary table were a joy to read !

They must have been written by someone at the local Take-Away, but after several attempts, I managed to make sense of them [I  am really a very bright lad ]

I must say the table works alright, the instructions just need to be a bit clearer.

[Adam Harris]

On Adam Harris Said:

Thanks for that further advice Bill. I have not figured out how to load images in a group into Google Images Translation rather than one by one  – maybe it only likes one at a time. I am using Windows 10

I’ve not tried this, and it might now be out of date, but possibly worth investigating; There’s probably even a Python user group where you could ask for assistance.

https://webapps.stackexchange.com/questions/171216/how-to-translate-images-with-google-translate-in-bulk

Bill

[Adam HarrisParticipant @adamharris13683]

Thanks Bill. Python etc beyond my flight range so I will plod on one page at a time. Its not actually really translation of image by google that is time consuming, but rather the other steps like scanning original and the careful checking of the final pdf output for manual editing where necessary. Whole process is taking about 5 minutes total per page which is surprisingly long but result is good and not too bad if spread over an hour or two each day.

[Adam HarrisParticipant @adamharris13683]

One commercial translator’s quote was over a thousand pounds!!

[Adam HarrisParticipant @adamharris13683]

Incidentally I am finding Google Images Translation very good on technical words for machinery, albeit not always perfect

[larry phelan 1]

On larry phelan 1 Said:

No Bill, I mean just trying to make sense of the “instructions” supplied with many of the machines and gear coming from China. The items work OK, but reading through the “guides” is an entertainment in itself.

The  “Instructions” which came with my rotary table were a joy to read !

They must have been written by someone at the local Take-Away, but after several attempts, I managed to make sense of them [I  am really a very bright lad ]

I must say the table works alright, the instructions just need to be a bit clearer.

In my experience, there are several factors at work explaining the poor translations of Chinese technical manuals:

The high costs Adam hints at for professional translation.

The technical vocabulary, requiring specialised knowledge of the subject matter by the translator, which is likely to be a rare occurrence when translating is the professional activity of the translator, not machining. This of course assumes the manufacturer has gone to the expense of employing a professional translator at all.

The great scarcity of native English translators working out of China, when it is almost a pre-requisite for producing a readable translation into another language to have it done by someone with native level proficiency in the target language.

Time I’ve spent in China has demonstrated to me that really the only faultless translations from Chinese into English you’ll find on display boards, signage or in information brochures etc. are the ones commissioned by central government, who will have access to the best translators, and these translations are typically on non-technical topics.

So what chance does the small provincial factory with tight margins have of producing a readable English language manual? Not much of one.

If anyone has a bilingual Chinese+English manual and they want clarification on what the Chinese actually says in a particular passage, I’m certainly willing to be of assistance.

 

[Graeme SeedParticipant @graemeseed34272]

Hi, on a comical side, a friend tried to translate “track rod end” into various languages. The Yanks call it different to us, Dutch call it knee joint etc etc. With a slight twist he went from english to spanish to dutch to japanese and many more in succesion and ended up with “can I have 5lbs of carrots please” Graeme.

[Author]

Posts