TXT → EPUB
Turn a plain text file into a real ebook — chapters detected, table of contents built, metadata set. Valid EPUB 3 that Kindle, Apple Books and Kobo all accept.
Drop your TXT
or click to choose — the only ceiling is your own memory, none on our side
A text file has words and nothing else. An ebook needs a manifest saying which documents exist, a spine saying what order to read them in, a navigation document for the table of contents, and metadata naming the book and its author. All of that has to be created, and most of it has to be guessed.
The guessing happens in two places. The encoding comes first: a text file carries no reliable statement of how its bytes map to characters, and books downloaded from older archives are very often Windows-1252 or an ISO-8859 variant rather than UTF-8. Decode one as the other and every accented character turns into a replacement glyph. This tool tries UTF-8 strictly first and falls back, and you can override it if you know better.
Chapter detection comes second. Lines that read like Chapter VII or Part Two start a new chapter by default; books that mark chapters with a line in capitals get a second rule; and if neither fits, one single chapter is a perfectly valid ebook. Check the chapter list in the preview before you send the file to your reader.
Format by format
| What it does | TXT | EPUB |
|---|---|---|
| What it stores | Characters and line breaks | Documents, a manifest, a spine, a nav document and metadata |
| Chapters | Implied by whatever the author typed | Real, listed in a table of contents |
| Title and author | Nowhere, unless typed into the text | In the OPF, where every reader reads them |
| Encoding | Not stated — has to be guessed | Declared as UTF-8 |
| Reading on a device | Awkward; no navigation | The format e-readers are built for |
| Accepted by stores | No | Yes |
What actually changes
Everything an ebook needs beyond the words has to be guessed or supplied. The guesses are good and you can override them, but check the chapter split before you send the result to your reader.
What survives
What does not
How it works
The encoding is sniffed — older files are often Windows-1252 rather than UTF-8.
Lines like “Chapter VII” start a new chapter. Change the rule if your book marks them differently.
With a nav document, an OPF and the mimetype entry stored first — the three things broken converters get wrong.
FAQ
The file is not UTF-8. Set the encoding option to Windows-1252 and convert again — that covers most older Western text files. ISO-8859-1 is the next thing to try. The preview shows which encoding was used, so you can check before downloading.
Switch the split rule. “On heading lines” looks for words like Chapter and Part; “On ALL CAPS lines” suits books that mark chapters that way; “Do not split” leaves the text whole, which is perfectly valid.
Yes. Amazon has accepted EPUB directly since 2022 and this output is valid EPUB 3, so Send to Kindle will take it. You no longer need to convert to MOBI first.
Not yet. The EPUB is valid without one and readers show a generated placeholder. Cover setting arrives with the next version of the metadata editor.
Yes — title, author and language are all fields on the form above, and they are written into the OPF where every reader looks for them. Leave the title blank and the file name is used.
Choose “Do not split” and you get a single-chapter ebook. That is a legitimate EPUB and reads perfectly well; you simply get no table of contents, because there is nothing to put in one.
Yes: version 3.0 in the OPF, a nav document flagged with properties="nav", a dcterms:modified timestamp, and the mimetype entry stored uncompressed as the first entry in the archive. Those last two are the ones most hand-rolled converters miss.
Once it is an EPUB, DualBook can turn it into a bilingual edition in over 30 languages.
