EPUBTurnPDF to EPUB converter

Find the cause of garbled EPUB text with a three-way check

By EPUBTurn · Updated

Compare the visible PDF phrase, copied PDF text and exported EPUB text to locate a recognition, encoding or reading-order problem before repairing it.

On this page
  1. Compare display and copied text
  2. Separate missing content from strange characters
  3. Never fill gaps with a plausible guess
  4. A useful three-way record
  5. Keep a short three-way text record
  6. Use a scan result to separate source and output

Compare display and copied text

Start with a short phrase you can read clearly in the PDF. Copy the phrase into a plain text editor and compare the characters. If the copied phrase is already wrong, the visible page and its extractable text disagree. If the copied phrase is correct but the EPUB looks wrong, look next at the output text and reader. That distinction prevents wasted changes to margins or font size.

Separate missing content from strange characters

Search the EPUB for an unusual word near the problem. Missing paragraphs, substituted characters and visual font problems can look similar on one screen but need different repairs. Check whether the full sentence is present and whether its words remain in order. A second reader can help distinguish an output problem from a display problem, but agreement between two readers is not proof that the source was preserved.

Never fill gaps with a plausible guess

A damaged name or numerical result is not an invitation to infer what the author probably meant. Keep the source and mark the uncertain passage. If you obtain a better authorised source edition, preserve the version difference. For scanned text, inspect the image against recognition output. When accuracy matters, use the PDF until the specific character or missing passage has been verified.

A useful three-way record

Write the visible PDF phrase, the phrase copied from the PDF and the phrase shown or copied from the EPUB. Keep the records short and specific. If the PDF image reads correctly but its copied text does not, you have identified a source extraction discrepancy rather than a generic reading-app failure.

If the output text contains the correct phrase but one reader displays missing symbols, record the reader and font setting before altering content. A second display test can help narrow the cause. Preserve the original character sequence until you know which stage is responsible, especially for names, units or non-English terms that should not be guessed from surrounding context.

Keep a short three-way text record

Download the text-layer diagnosis sheet to save the visible PDF phrase, copied PDF text and EPUB text together. Include a name, number or qualifier when the defect could change meaning.

Route a genuinely absent passage to missing-text checks, joined words to spacing repair, and a fused-looking letter pair to ligature diagnosis. Do not fix all three symptoms with the same global replacement.

Use a scan result to separate source and output

The historical cookbook excerpt already has imperfect copied text. The actual cookbook measures EPUB retains the checked measuring instructions and six rows as selectable text, making visible-source, copied-source and output comparison practical. Do not assign every bad source character to the converter. Check the exact phrase you need before deciding on another conversion.

EPUBTurn starts from the original PDF. Compare one phrase in the visible PDF, copied source text and sample EPUB.

Files used in this guide

Download the source files, conversion results or worksheets referenced here. Each label identifies the source document or the converter used.

1917 cookbook source excerptEPUBTurn actual measures sample: source pages 2–4

Related guides