Text is split at full stops, one pair per sentence.
Choose an EPUB file. Chapters and paragraphs will be detected automatically.
Paragraph — translate each paragraph as one block and display the same way. Fastest, fewest API calls.
Sentence — split every paragraph into sentences first, then translate each sentence on its own. The translator has less context, so pronouns and references can drift, but every row is exactly one source sentence next to its translation.
Sentence (aligned) — translate each paragraph as a single chunk (full context), then split both the source and the translation back into sentences and pair them up proportionally. Best of both: paragraph-level translation quality with sentence-level visual alignment.
Slow — 2 simultaneous translations. Easiest on Google’s free endpoint and never trips its rate limit, but takes the longest for big books.
Normal — 6 simultaneous translations. The balance that’s worked well in practice; the occasional 429 is retried with backoff.
Fast — 12 simultaneous translations. Often faster, but Google may start returning 429 (rate limit). The client automatically backs off and retries, so this can end up no faster than Normal — sometimes slower — on long books.
Preserve — treat each <br>-separated line as its own translation pair. Good for poetry, slogans, or any text where the line breaks carry meaning.
<br>
Collapse — ignore <br> inside paragraphs; the whole paragraph becomes one chunk. Useful for EPUBs that use <br> for visual line-wrap inside what is logically one paragraph.
Top 20 most-downloaded books on Project Gutenberg for the chosen language, ranked by download count.
No popular books found for "en". Check the language code.
Searches Project Gutenberg via gutendex. EPUB is preferred when available.