# PDF word count, scanned pages included A PDF word count reads the text of every page and adds up the words, characters and lines. Upload a PDF, DOCX, TXT or image below: a PDF with a text layer is counted on the spot, free and with no account, and scanned pages are read with OCR first. Each page is listed, so you can see which ones were scans. ## How to count the words in a PDF 1. **Choose the file.** A PDF, a Word file (.docx), a plain text file, or a photo or scan of a page (JPG, PNG), up to 5 MB. 2. **Press Count the words.** Pages with a text layer are counted in a second or two. Scanned pages go through OCR first, which takes a few seconds a page. 3. **Read the totals.** Words, characters with and without spaces, lines of 55 characters[1] and, when the text has any, Chinese, Japanese and Korean characters. 4. **Check the pages.** The page-by-page table shows each page's count and whether it was read from the text layer or with OCR. 5. **Price the job.** Type your rate per word or per line and your currency, and the quote line multiplies it out. Download the count as a CSV file for your records or your invoice. ## Why most PDF word counters miss scans A PDF exported from Word or a website carries a text layer, so any counter can read it. A scanned PDF is a picture of each page with no text behind it, and most online counters report zero words or skip those pages. Translators run into this on almost every certificate and contract. CAT tools such as memoQ don't extract text from scanned PDFs and tell you to run the file through an OCR program first[2]. On ProZ, translators have asked how to count a scanned PDF for well over a decade, and the usual answers are to count a few lines and extrapolate, or to OCR the file just to get a number[3]. This counter checks each page on its own. A page with a text layer is counted as it is, at no cost. A page that is only an image goes to OCR, and its count lands in the same table, marked OCR. Paid desktop tools such as AnyCount also count scans with built-in OCR, with licences from EUR 89 a year, on Windows only[4]. ## How each number is counted - **Words**: Runs of letters or digits. Don't, e-mail, 1,100.00 and 2026-11-02 each count as one word - **Characters with spaces**: Every character, with each run of spaces and line breaks counted as one space - **Characters without spaces**: The same text with every space left out - **Lines**: Characters with spaces divided by 55 and rounded up, the German Normzeile[1] - **CJK characters**: Chinese, Japanese and Korean characters, counted one by one Chinese and Japanese characters are not counted as words, because those languages don't put spaces between words, so a Chinese source is quoted per character. Korean puts spaces between words, so its words count as words and its characters are listed as well. Counters disagree about hyphenated words, numbers and text in headers or text boxes, so two tools rarely give exactly the same total. Expect a few words a page between tools, and a little more on a scan, where OCR can merge two words or split one. ## Counting lines for court and agency work German courts pay translators for every started line of 55 characters, the Normzeile, under § 11 JVEG[1]. The law counts the translation, not the source, unless only the source uses Latin script, so a count of the source is an estimate for the quote rather than the final bill[1]. It also pays a higher rate when the source did not arrive in an editable electronic form, which a scan never does[1]. The Kammergericht in Berlin counts characters with spaces electronically and divides by 55. Because a count by hand from a source that is not a Word file comes out higher, it accepts bills up to 20 percent above its own count[5]. Agencies and direct clients set their own unit: per source word, per target word, per line of 50 or 60 characters, or per page. Ask which one your client uses before you send the quote, then pick that unit in the quote line. ## What OCR gets wrong in a count OCR reads clean print well and handwriting badly. On a clear scan, the count usually lands within a few words a page of the original. Stamps, signatures, faint print and dense tables can add or drop words, so treat a scan's count as a close estimate and say so in your quote. The result page names every page that OCR read, so you know which numbers to double-check. Photos taken at an angle or in poor light count worse than a flat scan. If a page comes back with far fewer words than you expect, scan it again at a higher resolution and count it again. ## What is free, and when you need an account - **Text PDF, DOCX, TXT**: Counted free, with no account - **Scanned pages**: Read with OCR, up to 10 per file, 30 with a free account - **Before sign-up**: The total word count of a scan - **After sign-up**: Characters, lines, the page-by-page table and the CSV download - **Scans a day**: 2 without an account, 10 with one, shared with the other free tools - **Kept**: The file is deleted right after counting, the count after 1 hour Signing up takes an email address and a 6-digit code, with no password. It creates a free account, and counting does not use up the free document in the app. ## From a count to a translation Once the job is priced, the [LoreAI app](https://www.loreai.org/app) turns the scan into an editable Word file in the layout of the original, translated into one of 105 languages, ready to check or to load into your CAT tool. Your first document is free, up to 3 pages. To pull the text out of one page instead, use [PDF to text](https://www.loreai.org/pdf-to-text). For the signed statement that goes with a certified translation, use the [certificate of translation generator](https://www.loreai.org/certificate-of-translation-template). ## Questions people ask ### Can it count the words in a scanned PDF? Yes. Pages without a text layer are read with OCR, up to 10 per file without an account and 30 with a free one. The result says which pages were scans. ### Is this PDF word count free? Yes. A PDF with a text layer, a Word file or a text file is counted free with no account. For a scan you see the total words straight away and sign up free with your email to see the rest. ### Can it count words from an image? Yes. A JPG or PNG photo or scan of a page is read with OCR, like a scanned PDF page. ### How many characters make a line? The counter uses 55 characters including spaces and counts every started line, as German courts do[1]. If your client uses another line length, divide the characters with spaces by that number. ### Is my file kept? No. The file is deleted as soon as it is counted. Scanned pages go to LoreAI's OCR provider and are deleted there right after reading. The count is kept for 1 hour so you can open it after signing up. ## Sources 1. [Gesetze im Internet: § 11 JVEG, Honorar für Übersetzungen](https://www.gesetze-im-internet.de/jveg/__11.html), accessed November 2, 2026 2. [memoQ documentation: PDF (Portable Document Format) files](https://docs.memoq.com/current/en/Workspace/pdf-portable-document-format-files.html), accessed November 2, 2026 3. [ProZ.com forum: Word count for scanned PDF files](https://www.proz.com/forum/money_matters/172658-word_count_for_scanned_pdf_files.html), accessed November 2, 2026 4. [Translation3000: AnyCount word count tool](https://www.translation3000.com/word-count-tool/), accessed November 2, 2026 5. [Kammergericht Berlin: Hinweise zur Vergütung von Übersetzungen (10/2019)](https://www.berlin.de/gerichte/_assets/was-moechten-sie-erledigen/hinweise_verguetung_uebersetzungen_8_10_2019.pdf), accessed November 2, 2026 By Matt Mountain · Updated November 2, 2026