What are full-width and half-width characters?

Full-width (zenkaku) characters occupy the same square cell as a kanji, while half-width (hankaku) characters take half that width. Japanese text often mixes full-width Latin letters and digits with normal half-width ASCII, which causes inconsistent data.

How does the ASCII conversion work?

Full-width ASCII lives in the Unicode block U+FF01 to U+FF5E, which sits exactly 0xFEE0 above the matching plain ASCII codes U+0021 to U+007E. Converting is a fixed code-point shift in either direction, with the full-width space U+3000 mapped to and from a normal space.

What about half-width katakana?

When converting to full-width you can also map half-width katakana (U+FF61 to U+FF9F) to standard full-width katakana. The tool recombines a half-width dakuten or handakuten that follows a base kana, so ｶﾞ becomes ガ and ﾊﾟ becomes パ.

Why normalise width before storing data?

Search, sorting, and de-duplication treat Ａ and A as different characters, so width inconsistency causes missed matches and duplicate records. Normalising ASCII to half-width and katakana to full-width gives a single canonical form for indexing.

No. All conversion happens in your browser using code-point arithmetic and a built-in katakana table. Nothing you paste is sent to a server.

What is the Japanese Full-Width / Half-Width Converter?

Converts full-width ASCII letters, digits and punctuation (Ａ-Ｚ, ０-９) to half-width and back, with correct half-width katakana conversion including recombining dakuten and handakuten marks. It runs free in your browser on Gera Tools, with nothing uploaded.

Japanese Full-Width / Half-Width Converter

Name: Japanese Full-Width / Half-Width Converter
Creator: Gera Tools
License: https://creativecommons.org/licenses/by/4.0/

Japanese text frequently mixes full-width (zenkaku) and half-width (hankaku) characters, which breaks search and data consistency. This converter normalises ASCII and katakana between the two widths using the correct Unicode mappings.

Why full-width and half-width coexist in Japanese text

Traditional Japanese typesetting allocated a fixed square cell to every character — kanji, hiragana, katakana all occupy one square. When ASCII was incorporated into Japanese character sets, two versions emerged: a half-width form that took half a cell (matching standard ASCII), and a full-width form that took a full cell, fitting neatly in mixed Japanese-ASCII layouts.

Over decades, both versions ended up encoded in Unicode:

Half-width ASCII is the familiar U+0021..U+007E range — the ordinary keyboard characters.
Full-width ASCII is a parallel block at U+FF01..U+FF5E — visually wider, as in Ａ, Ｂ, １, ２.
Half-width katakana lives at U+FF61..U+FF9F — a legacy encoding used in older systems like JIS X 0201.
Full-width katakana is the standard modern form in the U+30A0..U+30FF block.

Text from legacy databases, OCR output, or older Japanese web pages often mixes these freely. When a user types Ａ and your database stores A, a search for one misses the other entirely.

How the conversion works

Full-width ASCII is a clean offset of plain ASCII, so the conversion is arithmetic rather than a lookup table:

full-width ASCII  U+FF01..U+FF5E  =  ASCII U+0021..U+007E + 0xFEE0
full-width space  U+3000          <-> half-width space U+0020
half-width katakana U+FF61..U+FF9F -> full-width katakana (table)
   ｶ + ﾞ (dakuten)    -> ガ
   ﾊ + ﾟ (handakuten) -> パ

To go from full-width to half-width, subtract 0xFEE0; to go the other way, add it. For half-width katakana the tool uses a mapping table and looks ahead one character to recombine a trailing voiced or semi-voiced mark into the single composed kana.

Dakuten and handakuten: the tricky part

Half-width katakana represents voiced sounds (like ガ) as two separate characters: the base kana (ｶ) followed by a combining mark (ﾞ). When converting to full-width, these two characters must be merged into a single precomposed code point (ガ). If you skip this step and convert each code point independently, you end up with カ゛ — two characters that look like the voiced kana but are not the same Unicode character, causing subtle mismatches in search and sort.

The tool handles this by peeking at the character after each base kana and consuming the combining mark if it is present.

Practical examples

Input	Direction	Output	Notes
`Ｈｅｌｌｏ　２０２６`	Full→Half	`Hello 2026`	Full-width letters and space
`ｶﾞｲﾄﾞ`	Half→Full	`ガイド`	Dakuten recombined
`ﾃｽﾄ`	Half→Full	`テスト`	Plain half-width katakana
`Hello`	Half→Full	`Ｈｅｌｌｏ`	Standard ASCII widened

When to use each direction

Normalise to half-width ASCII + full-width katakana before indexing user input or storing in a database. This gives a single canonical form so that Ａ and A match, and ﾃｽﾄ and テスト match.

Normalise to full-width when preparing text for display in a Japanese context where consistent character-cell width matters — for example, generating fixed-width terminal output or preparing text for a legacy system that expects zenkaku.

Apply the same normalisation to both the stored data and the search query to ensure symmetric matching.