How is this different from Google Translate or DeepL?

Google and DeepL translate sentence-by-sentence without understanding that fiction has genre conventions, character voice, dialogue punctuation rules, and chapter structure. Our engine applies native dialogue formatting (guillemets for French and Italian, German low-quotes „like this", em-dashes for Spanish and Russian), preserves proper names and made-up terms, handles scene breaks, and keeps your character voices consistent across 50+ languages.

Which languages does it support?

Spanish, French, German, Italian, Portuguese, Japanese, Chinese (Simplified), Russian, and 40+ others with varying levels of dialogue-formatting awareness. Spanish, French, German, and Italian have the most mature dialogue rules built in. Others use standard quotation marks with full voice preservation.

Can I translate my whole manuscript?

The free sample tool handles 2,500 words a run — enough for a chapter opening or a critical scene you want to test. For the full manuscript, use the paid Translation Service: we chunk your book chapter-by-chapter, build a story bible first for character consistency, then translate with your chosen model.

Will it preserve my character voices?

Yes. The prompt explicitly preserves proper names, made-up terms, magic-system vocabulary, and character-specific speech patterns. It also maintains narrative perspective (first person stays first person) and tense consistency. Idioms get adapted to sound natural in the target language without flattening your original voice.

Should I run editing first or translate first?

Always edit first. Run your English draft through grammar-check, slop-scanner, and the rewriter before translating. A clean source yields a dramatically better translation — garbage in, garbage out. The translator amplifies whatever prose quality you feed it.

MULTILINGUAL FICTION ENGINE

Don't Settle For
KDP Auto-Translation.

Fiction needs nuance, pacing, emotional intelligence. DeepL produces robotic sentences. We preserve your voice across 9+ languages with native dialogue formatting.

🌍 Target Language

🔧 Engine

Paste chapter to translate: 0 / 2,500 words

Free up to 2,500 words a run — bring your whole manuscript to the paid tools.

📖 Full Manuscript Context

We chunk your manuscript elegantly, preserving scene continuity. Before translation begins, the AI builds a Story Bible so it understands your protagonist's voice, the supporting cast, and your world's vocabulary.

⚡ Native Polish Pass

After the literal translation, run a secondary line-edit pass in the target language. Tell the engine to "Line Edit" the Spanish version to elevate the prose, fix clichés, and restore the gripping cadence of your original work.

Why Fiction Translation Is Different From Regular Translation

Machine translation of nonfiction is largely a solved problem. Google Translate can render a news article, a technical manual, or a business email into Spanish and the result will be functional. Fiction is different. Fiction has conventions that don't travel across languages — and the machine translators of the world ignore them.

The biggest one: dialogue formatting. American English uses straight quotation marks. Spanish uses em-dashes for dialogue. French and Italian use guillemets (« »). German uses low-quote/high-quote („like this"). If you translate a Spanish novel into German and leave the dialogue in em-dashes, a German reader knows instantly that the book was badly formatted — before they read a single word of prose. That's one lost sale every time. DeepL doesn't fix this. KDP auto-translate doesn't fix this. Our engine does.

What The Engine Handles Natively

Dialogue formatting per language: em-dashes for Spanish and Russian, « guillemets » for French and Italian, „German low-quotes", standard marks for Portuguese, Japanese, and Chinese. Applied automatically based on target language.
Chapter headings: "Chapter 1: The Beginning" in French becomes "Chapitre 1 : Le Début" (with the correct French spacing rule around the colon). "Prologue" → "Prólogo" (Spanish), "Prolog" (German), "Prologo" (Italian).
Scene breaks: a line containing only "***" is preserved as a scene break on its own line.
Proper names: character names, place names, and made-up terms stay in the original. Harry Potter is still Harry Potter in French.
Magic systems and invented vocabulary: consistency is enforced across chapters. If you named a spell "Windrising" in chapter 1, it stays "Windrising" in chapter 47.
Markdown preservation: italics stay italics, bold stays bold, paragraph structure is preserved exactly.
Voice preservation: if your narrator is snarky in English, they're snarky in German. The prompt explicitly asks the model to match the tone and pace of the original, not flatten it toward a neutral translation voice.
No meta-commentary: the output is the translation only. No "Here is the translated text:" preamble, no "This passage was challenging because" notes. Just the prose, ready to paste into the target-language manuscript.

Best Practices

1. Proofread your English draft first

A clean source yields a dramatically better translation. Run your English manuscript through Grammar Checker, Slop Scanner, and The Rewriter before you translate. Every mistake in the English version gets amplified in translation.

2. Test one chapter before committing

Before dropping credits on a 90,000-word full book, translate chapter 1. Have a native speaker or beta reader check the tone. Every language has its own commercial fiction register — make sure you're landing in the right one before you translate the whole book.

3. Choose your engine deliberately

Gemini 3.1 Flash-Lite is fine for rapid drafts and short excerpts. For a full manuscript you care about, use a premium model — GPT-5.4 High, Claude 4.5 Sonnet, or Gemini 3.1 Pro. The price difference is real but so is the quality difference. Fiction translation is one place where the premium models actually matter.

4. Budget for a native pass

No matter how good the engine, hire a native speaker to skim the final output before publishing. They'll catch cultural references that don't land, idioms that feel stilted, and the occasional phrase that reads wrong to a native ear. The machine does 95%. The human closes the last 5%.

Related Tools

Grammar Checker — clean your English source before translating.
The Rewriter — polish your English prose before it gets translated.
Slop Scanner — strip AI tells before the translator amplifies them.
Full Editing Services — for the complete source-preparation pipeline.

Don't Settle For KDP Auto-Translation.

Translate The Whole Book