Skip to content
Account

Free sample files for testing

10 fictional documents to download and test with — scanned pages with no text layer, an invoice and its scan, a bank statement, a receipt, a fillable form and a notice in three languages. 15 files in all, each under 500 KB, every one marked SPECIMEN and free to use under CC0.

Sample scanned PDF for OCR testingOne page, image only, 200 dpi, skewed — the basic OCR test.1 page · PDF 294 KB · image only, no text layer · 200 dpi · CC0Sample multi-page scanned PDFThree scanned pages, each skewed differently, with two tables.3 pages · PDF 369 KB · image only, no text layer · 150 dpi · CC0Sample scanned letterA colour scan of a one-page letter, folded, with a date and an amount.1 page · PDF 198 KB · image only, no text layer · 150 dpi · CC0Sample invoice PDFA digital invoice with six lines, GST and totals that add up.1 page · PDF 30 KB · real text layer · CC0Sample scanned invoiceThe sample invoice as a scan — same numbers, no text layer.1 page · PDF 270 KB · image only, no text layer · 200 dpi · CC0Sample bank statement PDFTwo pages from "Specimen Bank", 32 transactions, marked SPECIMEN.2 pages · PDF 33 KB · real text layer · CC0Sample receiptAn 80 mm till receipt, as a PDF and as a phone photo.1 page · PDF 36 KB · real text layer · CC0Sample fillable PDF formA one-page application form with text, radio, checkbox and drop-down fields.1 page · PDF 47 KB · real text layer · CC0Sample PDF for translation testingOne notice, three languages, one page each — with a reference translation.3 pages · PDF 33 KB · real text layer · CC0Sample scanned rental agreementA worn two-page agreement scan with a coffee ring, a fold and two signatures.2 pages · PDF 385 KB · image only, no text layer · 150 dpi · CC0

Every file, with its size

What these files are

Test documents that behave like the real thing. The scans were drawn, rendered to pixels and saved with no text layer, then given the faults real scans have: a few degrees of skew, grain, dust, a fold, a darker corner. The digital files have real text, and their numbers add up — every invoice line is quantity times price, and the bank statement's running balance is right on every row — so what a tool extracts can be checked against the page.

None of it is real. The companies, people, addresses, account numbers and amounts are invented, and every page says SPECIMEN. All of the files are dedicated to the public domain under CC0 1.0.

Looking for sample CSV, Excel or Parquet data instead? Our sister site ExploreMyData has free sample datasets.