How to batch translate scanned documents and process multiple files at once

Key points of this article

There are actually two types of batch processing: submitting them one by one, or merging them into one and then translating them as a whole. First determine whether the files are independent, and then choose a method.
Before submitting each document, you must confirm that it is a scanned copy by trying to select text. If you fail to select, it will basically be a picture-type scanned copy.
Immediately rename by number and register it to the list after downloading, which is more reliable than recalling which copy corresponds to which copy later.
Before the translation is delivered, check the number of pages, key figures, dates and names of people item by item. Machine translation may make mistakes, and the original shall prevail.

If you have a stack of scanned documents in your hand, such as a dozen scanned contracts, dozens of pages of paper handouts, or multiple scanned papers, you most likely want to translate them all at once. But "batch" here actually includes two things: first, lining up multiple documents and sending them for translation one by one; second, merging pages belonging to the same material and then translating them again. The order of operations and verification methods of the two are different, and if they are mixed together, it is easy to miss parts or submit them twice.

This article is for students, paralegals, administrators, and researchers who need to deal with multiple scans. After reading, you will know: how to determine whether each document is a scanned copy before taking action; how to submit one by one on OmniDoc and how to organize the output; when to merge or split first; how to check each result before delivery. It should be noted that the approach here does not rely on any batch entry, and each document is submitted individually.

How to batch translate scanned documents? One sentence conclusion

First determine whether these files are independent of each other. Independent scans are uploaded one by one, downloaded one by one, and named uniformly; the same material is scanned into many small files, first merged into one PDF and then translated again, so that the terminology and translation style are easier to maintain consistency.

your situation	Recommended practices	reason
Multiple independent documents, such as several contracts, several papers	Upload the translations one by one and archive them by number after downloading	Each result is independent, making it easy to locate errors.
The same material is split into multiple scanned files, such as handouts scanned by chapters.	Merge into one PDF first, then translate as a whole	The terminology is the same, and the verification only needs to be done once
There are many pages in a single document, making it uneasy to submit it at once	First break it into several sections, translate them into sections and then put them back together according to page numbers.	If there is a problem with a single section, you only need to redo that section.
The page is blurry, skewed, or contains many handwritten comments	Rescan first, or process it with OCR on desktop client before submitting	The text is clearer, making it easier to check against the original text
The file formats are mixed, including scanned PDF, Word or pictures.	Group by format and process similar items together	The output formats are different and are less likely to be confused after grouping.

Before you start: Confirm these points first

1. Confirm that each file is indeed a scanned copy. Open the PDF, try to select a piece of text with the mouse and copy it. If the selection cannot be selected, or the copy is blank, it can basically be judged as a picture-type scanned copy. This step must be done one by one, because some PDFs have both text layers and pictures, which is different from pure pictures and should be marked separately when grouping.

2. Count the number of pages and file size. Before submitting, note down the total number of pages of each document as a baseline for subsequent verification. If a commit fails, see if it's significantly larger than you're comfortable with before splitting it instead of trying again.

3. Decide the output method: bilingual comparison or translation only. OmniDoc's document translation supports bilingual side-by-side or translation-only output. For contracts and papers that need to be checked paragraph by paragraph, it is recommended to choose bilingual comparison to facilitate comparison of the original text; if you only plan to read the general meaning of the lecture notes, only the translation will be enough.

4. Select style according to document type. There are different style options for academic, legal, medical, technical, business, literary and other texts. When a batch of file types are mixed, grouping them by type and selecting styles separately is more secure than using the same settings for the entire batch.

5. Specify naming and archiving rules in advance. For example, the format of "number_original file name_language". Start by setting the rules, and once you download them, you don’t have to guess which piece corresponds to which piece.

Complete the steps to translate scanned documents in OmniDoc one by one

Group and number by materials. Divide the documents into groups and number each document in the list, such as "01 Contract A" "02 Contract B" and note the number of pages.
Open omnidoc.top, enter document translation, and upload the first scanned PDF. Scanned PDFs can be uploaded directly without additional processing first.
Select the output method and style. Select bilingual comparison or translation only according to the decision in the first step, and select the style corresponding to the document type.
Submit and wait for the task to be completed and download the results. Do not close the page after downloading, and make sure the result file can be opened normally.
Rename by number immediately. For example, "01_Contract A_Translation.pdf" should remain consistent with the original file name to avoid confusion later.
Tick off the list before working on the next one. Register every time you complete one to prevent missing work or repeated submissions.
If you encounter abnormal results, check back first, do not overwrite them directly. When there are few pages, blank pages or garbled characters, go back to the original file to check whether there are any problems with the scan quality, split or resubmit if necessary, and keep the confirmed version.
Arrange and archive after everything is completed. Put the translation and the statistical table in the same folder. The statistical table at least contains the file name, page number, output method and verification status.

If several documents belong to the same material and need to be merged first, you can use the merge or split function in the page operation in the PDF editor of the desktop client, and then submit according to the above steps. Before merging, be sure to write down the page number range corresponding to each document, and then split it back according to the range.

Check each item after completion

Check item	What do you think?	What if there's a problem?
Is the number of pages consistent?	Compare the number of pages between the original scan and the translation	Check back if it is missing when there are fewer pages, or if you miss a paragraph when you split it.
Is there any blank or missing picture on the page?	Skim through the pages page by page, focusing on the tables, signatures, and charts	Resubmit the copy and retain the original results for comparison
Whether it is bilingual or not	Spot check the beginning, middle and end paragraphs to see if the source text corresponds to the translation one by one.	Check the page number order if it doesn't match, and resubmit it if necessary
Key number, date, person name, amount	Item by item comparison with the original scan	Both machine translation and OCR may make mistakes, which shall be subject to the original and corrected manually
Whether the same term is unified or not	Search multiple translations for the same term to see if the translation is consistent	Documents that merge and translate can first unify the glossary and then check
Whether the file number corresponds to the list one by one	Compare file names with list numbers one by one	Fill in the missing items, delete duplicates, and then file

# Core comparison

Method	An appropriate situation	Advantages	Need to pay attention to
Submit one by one	Files are independent of each other	The result is independent, and the error is easy to locate.	Documents should be patiently registered for a long time.
Merge first and then translate as a whole	The same material is broken down into multiple files	Consistency of terminology and style	Write down the page number range before merging to facilitate removal
First split and then segment translation	There are many pages in a single file.	Each paragraph is short, so it is easy to check paragraph by paragraph.	Check the continuity of the page number when splicing

# which method is suitable for you?

-for the processing of contracts, supporting materials and other documents that need to be cross-referenced one by one, it is recommended that they be submitted one by one and bilingual comparison should be selected, with emphasis on figures and terms. -organize the course handouts or multi-chapter scanned copies of the same book. It is recommended to merge them by chapter and then translate them as a whole to ensure the unity of terminology. -in the face of a scanned book with a large number of pages, it is recommended that it be divided into several paragraphs, submitted in segments, and checked separately after each paragraph is completed. When you only need to read through the main idea, you can choose only the translation, but the content related to the decision still has to go back to the original for confirmation.

# frequently asked questions (FAQ)

1. Can scanned pieces and ordinary PDF be translated in batch together?

It can be sorted out together, but grouping is recommended. The key points of the two checks are different, and after grouping, they can be checked group by group according to the same standard, and it is easier to judge whether the result is normal or not.

2. What if the scanned piece is skewed and blurred?

First, let's see if the original can clearly identify the text. It is usually easier to rescan a blurred or skewed page and then submit it than to repair it afterwards. You can also OCR the scanned piece on the desktop client first, and then check whether the text can be selected properly.

3. After the merger and then translation, can the result be taken back according to the original file?

Yes, as long as the page number range of each document is recorded before the merger. Switch back according to these ranges when splitting, and then check with the number of pages in the list one by one.

4. There are so many files, is there any way not to upload them one by one?

The practice of this article does not rely on batch imports and needs to be submitted one by one. When there are many documents, they can be completed in several times, each time controlled in the number that can be fully checked, and the progress can be recorded in the list.

# Summary

How to do the batch translation of scanned pieces, the key is to judge the relationship between the documents first: submit them one by one independently, merge the same material first and then translate it as a whole. Confirm the scan type before operation, file according to the number after download, check the number of pages, bilingual correspondence and key information item by item before delivery, and multiple documents can be completed in an organized manner.

Open OmniDoc, upload the document and start translation.