Vertical Japanese Translator
Vertical Japanese is where general translation tools fail most reliably. Characters run top to bottom and right to left, and reading them with horizontal assumptions produces sentences in scrambled order — every word recognisable, the whole meaningless. Old books and periodicals set vertically have the same problem.
Upload a vertical-text document to start translating
Geyi handles vertical layouts specifically: recognition follows the vertical reading order, two-page spreads and multi-column pages are corrected automatically, and the translation comes back set horizontally.
Geyi handles vertical layouts specifically: recognition follows the vertical reading order, two-page spreads and multi-column pages are corrected automatically, and the translation comes back set horizontally. It covers vertical Japanese and vertical Chinese.
Key capabilities
- Vertical Japanese and vertical Chinese, output horizontally
- Two-page spreads restored as one reading order
- Multi-column pages restored column by column
- Captions and marginal notes kept with their layout
- Document extraction returns the original text without translating
Layouts handled automatically
- Vertical Japanese and vertical Chinese: recognised and output horizontally.
- Two-page spreads: restored as one reading order across the spread, with no need to split pages yourself.
- Multi-column pages: restored column by column rather than swept across.
- Captions and marginal notes: kept with the layout they belong to rather than dropped into the body text.
Extract first, or translate directly
If you need the original text out of a classical work for research, use document extraction: 4 credits a page, extraction only, output in the original language as Word, PDF, TXT or Markdown.
If you need to read the content, go straight to file translation on AI reflow, which recognises and translates in one pass. When you are unsure about recognition quality, extracting one page first is the cheapest way to find out.
Common obstacles with old material
Reproduction quality decides most of it: clean print scanned straight is recognised best, while yellowed paper, ink bleed and skewed scans all make it harder. Variant and archaic character forms may be normalised to modern shapes.
For textual scholarship, treat the translation as a reading aid and take key conclusions from the original.
Geyi compared with generic machine translation and a manual workflow
The generic machine translation and manual workflow columns describe common cases; actual results vary by tool and method. The Geyi column lists only capabilities that are live in the product today.
See it in action
Four opening pages of the 1917 Rashōmon scan are translated through high-precision reflow and then reviewed paragraph by paragraph as horizontal Simplified Chinese. Furigana leaking into prose, clothing and time references, and page-break joins are corrected for display.
Vertical-book translation: page 1 of the 1917 edition and the reviewed opening paragraph in Simplified Chinese, without furigana mixed into the prose.
Printed pages 1–4, arranged in Japanese reading order, alongside the complete reviewed excerpt with text joined across page breaks.
Before
或日の暮方の事である。一人の下人が、羅生門の下で雨やみを待つてゐた。廣い門の下には、この男の外に誰もゐない。唯、所々丹塗の剝げた、大きな圓柱に、蟋蟀が一匹とまつてゐる。羅生門が、朱雀大路にある以上は、この男の外にも、雨やみをする市女笠や揉烏帽子が、もう二三人はありさうなものである。それが、この男の外には誰もゐない。
After
某日黄昏时分，一个下人正在罗生门下等待雨停。宽阔的门下，除了这个男人以外谁也没有。只有在那几根各处剥落着朱红色油漆的大圆柱上，停着一只蟋蟀。既然罗生门位于朱雀大路上，除了这个男人之外，原本似乎还应有两三个戴着市女笠或揉乌帽子、在此躲雨的人。然而，除了这个男人以外，谁也没有。
Opening paragraph: transcribed source text with old-style characters and furigana omitted, alongside the manually reviewed Simplified Chinese translation.
Processed through Geyi’s actual pipeline; the translation is manually checked and edited paragraph by paragraph. The display is an edited excerpt. · Source: Rashōmon, Ryūnosuke Akutagawa, 1917 (The author, Ryūnosuke Akutagawa, died in 1927, more than 70 years ago; under Japanese copyright law, works of authors who died before the end of 1967 were protected for 50 years after death, so protection of this book expired at the end of 1977. The National Diet Library Digital Collections marks the item PDM (Public Domain Mark) and publishes it online (DOI 10.11501/1088339).)
Frequently asked questions
Will vertical text come back scrambled?
No. Recognition follows the vertical reading order from the start, and spreads and multi-column pages are corrected automatically.
Is the translation still vertical?
No — output is horizontal throughout, which reads better on screen and is easier to edit afterwards.
How is it billed?
Vertically set originals are usually scans, so they take the recognition path used by AI reflow: 8 credits a page on high precision, 4 on standard, with a 10-credit minimum per file.
Can I get the original text without translation?
Yes — use document extraction at 4 credits a page, which extracts without translating.
Are archaic character forms recognised?
Most are, though they may be normalised to modern forms. For textual research, work from the original.
Does this work for classical Chinese texts?
Yes. Vertical Chinese takes the same recognition path as vertical Japanese, with the same reading-order correction.
Related
简体中文 · Français · Deutsch · Русский · 日本語 · 한국어 · العربية · 繁體中文
Geyi is a proprietary AI document platform owned and operated by Nayuta Technology and Culture Company Limited.