How do I type IPA symbols?

Put the cursor where the symbol should go, then click its key: the board has tabs for diacritics, consonants, vowels, other symbols and suprasegmentals. You can also search by Unicode name, by sound (fricative, rounded), by nickname (ash, wedge) or by U+ code and press Enter to insert the first match, or type an X-SAMPA code into the shortcut box. Copy puts the text on your clipboard and Save .txt downloads it as a UTF-8 file.

What is X-SAMPA, and how do the shortcuts work?

X-SAMPA spells IPA symbols in plain ASCII. The code under each key is its X-SAMPA code, and codes are case-sensitive: s is s, S is ʃ, E is ɛ and @ is ə. The box takes one code or a whole transcription, so /h@"loU/ becomes /həˈloʊ/. Where two readings are possible the longest code wins, a hyphen splits two codes, and a lone underscore is the tie bar, so t_S gives t͡ʃ. Tone letters and a few marks have no X-SAMPA code and come from the board or from search.

How do I add a diacritic, like the nasal tilde over a vowel?

Type or insert the vowel, leave the cursor right after it, then click the tilde key (or type ~ in the shortcut box). The mark attaches to that vowel, and further marks stack on the same symbol, so ɛ plus the tilde is one character made of two code points. If the cursor is at the start of the text or after a space, the mark is refused with a message instead of being inserted on nothing. To show a mark on its own, insert the dotted circle ◌ first.

Can it transcribe a word into IPA or tell me how to pronounce it?

No. This is a keyboard: it has no dictionary, no speech input and no rules for any language, so it cannot tell you how a word is said. You choose the symbols and it inserts exactly the characters you chose.

Why does a symbol look different, or show as a box, after I paste it?

The code point is exact, but how it is drawn depends on the fonts where you paste it. Symbols from newer parts of Unicode, such as U+1DF04, and stacked diacritics need a font that covers them. The code point table on this page lists your text code point by code point, up to the first 300, with the UTF-8 bytes and the Unicode name of every symbol on the board, so you can check what is really there.

Is the ɡ on this keyboard the same as the g on my keyboard?

No. The key for the voiced velar plosive inserts U+0261 LATIN SMALL LETTER SCRIPT G, the letter the IPA uses, while your keyboard types U+0067 LATIN SMALL LETTER G. Many fonts draw them alike, but they are different characters, so a text search for one will not find the other. The X-SAMPA code g gives U+0261.

Why are code points and characters counted separately?

A symbol with a diacritic is one character on screen but two or more code points, and a character limit may count either. The readout shows both, plus the size in UTF-8 bytes: /ʃɪp/ /vɛ̃/ is 11 code points, 10 characters and 15 bytes.

Is my text uploaded or saved?

Nothing is uploaded: the page works on your text in the browser. Your draft and the tab you had open are kept in this browser's local storage so they are there next time, and the Start over button forgets them.

IPA Keyboard

Every symbol on the official IPA chart on one board: consonants, vowels, other symbols, diacritics, and stress, length and tone marks, each key labelled with what it stands for and its exact Unicode 17.0 name. Click a key and the symbol lands at your cursor, find one by searching for esh, fricative or U+0283, or type X-SAMPA into the shortcut box, where S gives ʃ and E~ gives ɛ̃. A combining mark attaches to the symbol just before the cursor, and one with nothing to sit on is refused with the reason instead of left hanging in the text. Undo and redo cover your typing and every insert, and the counts under the box show code points, characters and UTF-8 bytes as you go.

Your transcription

0 code points, 0 characters, 0 bytes as UTF-8; nothing typed yet

Search names or U+ codes

Shortcut, then Enter (X-SAMPA)

Codes are case-sensitive: S is ʃ, E is ɛ, ~ nasalizes.

Tab to the board, then arrows move between keys, Enter inserts, / searches, Ctrl+Z undoes.

Code points

Characters

UTF-8

0 bytes

Names

Unicode 17.0

Every key inserts one code point at the cursor, or in place of what you selected. A combining mark such as the tilde joins the symbol just before the cursor, and a mark with nothing to sit on is refused with the reason. Search finds a key by its Unicode name, what it stands for, a nickname such as ash or wedge, its U+ code or its X-SAMPA code; paste a symbol into search to find out what it is. Your own keyboard types into the box as usual.

What comes out

Select U+0283from the board, or shortcut S
ʃʃ inserted at the cursor
U+0303 after ɛcombining tilde, cursor after U+025B
ɛ̃ɛ̃: the tilde stays on ɛ, 2 code points, 1 character
Replace, then undoə typed over ɪ selected between two U+1DF04
𝼄ə𝼄both U+1DF04 kept whole; undo brings back ɪ

How the shortcuts read

The shortcut box takes X-SAMPA, an ASCII spelling of the IPA, one code or a whole transcription at a time. The code under each key is its X-SAMPA code. Where X-SAMPA could be read two ways, this is how the box reads it.

Codes are case-sensitive: s gives s and S gives ʃ; e gives e and E gives ɛ.
The longest code at each position wins: r\` is ɻ, @` is ɚ, |\|\ is ǁ, and _H_T is the one high rising tone mark rather than a high mark and an extra-high one. A hyphen splits two codes and types nothing, so r\-` gives ɹ˞ and _H-_T gives the two marks.
An underscore followed by a diacritic letter is that diacritic (_h is ʰ, _0 is the voiceless ring). On its own it is the tie bar, so t_S gives t͡ʃ. That makes k_x read as k with the mid-centralized mark; to tie k and x, use the tie-bar key.
The apostrophe is palatalization (ʲ), not stress. Primary stress is " (ˈ) and secondary stress is % (ˌ).
g gives ɡ, U+0261 LATIN SMALL LETTER SCRIPT G, the IPA letter, rather than the g on your keyboard (U+0067).
Some symbols have two codes: ʲ is ' or _j, nasalized is ~ or _~, syllabic is = or _=, and ʋ is P or v\.
Tone letters, the labiodental flap ⱱ, the ties below and a few marks have no X-SAMPA code: choose them from the board or find them by name.
Spaces, slashes, square brackets and parentheses pass through unchanged, so /h@"loU/ gives /həˈloʊ/.
Unicode names are kept exactly, even where they are historical: ǃ is named LATIN LETTER RETROFLEX CLICK and ǂ LATIN LETTER ALVEOLAR CLICK, though the chart uses them for the alveolar and palatoalveolar clicks. Each key shows both.

Code points in your text

Type or insert something and each code point is listed here with its name and its UTF-8 bytes.

Exact about code points: Character entry only: each key inserts the one Unicode code point it names, labelled with its Unicode 17.0 name, and your text copies as plain text and saves as a UTF-8 .txt file. It does not transcribe speech or decide how any word is pronounced. A combining mark such as U+0303 attaches to the symbol just before the cursor, shortcut codes follow X-SAMPA, and how a symbol looks depends on the fonts where the text is shown. The text stays in this page, so nothing is uploaded. Your draft and the open tab are kept in this browser, and the Start over button above forgets them.

Common questions

How do I type IPA symbols?: Put the cursor where the symbol should go, then click its key: the board has tabs for diacritics, consonants, vowels, other symbols and suprasegmentals. You can also search by Unicode name, by sound (fricative, rounded), by nickname (ash, wedge) or by U+ code and press Enter to insert the first match, or type an X-SAMPA code into the shortcut box. Copy puts the text on your clipboard and Save .txt downloads it as a UTF-8 file.
What is X-SAMPA, and how do the shortcuts work?: X-SAMPA spells IPA symbols in plain ASCII. The code under each key is its X-SAMPA code, and codes are case-sensitive: s is s, S is ʃ, E is ɛ and @ is ə. The box takes one code or a whole transcription, so /h@"loU/ becomes /həˈloʊ/. Where two readings are possible the longest code wins, a hyphen splits two codes, and a lone underscore is the tie bar, so t_S gives t͡ʃ. Tone letters and a few marks have no X-SAMPA code and come from the board or from search.
How do I add a diacritic, like the nasal tilde over a vowel?: Type or insert the vowel, leave the cursor right after it, then click the tilde key (or type ~ in the shortcut box). The mark attaches to that vowel, and further marks stack on the same symbol, so ɛ plus the tilde is one character made of two code points. If the cursor is at the start of the text or after a space, the mark is refused with a message instead of being inserted on nothing. To show a mark on its own, insert the dotted circle ◌ first.
Can it transcribe a word into IPA or tell me how to pronounce it?: No. This is a keyboard: it has no dictionary, no speech input and no rules for any language, so it cannot tell you how a word is said. You choose the symbols and it inserts exactly the characters you chose.
Why does a symbol look different, or show as a box, after I paste it?: The code point is exact, but how it is drawn depends on the fonts where you paste it. Symbols from newer parts of Unicode, such as U+1DF04, and stacked diacritics need a font that covers them. The code point table on this page lists your text code point by code point, up to the first 300, with the UTF-8 bytes and the Unicode name of every symbol on the board, so you can check what is really there.
Is the ɡ on this keyboard the same as the g on my keyboard?: No. The key for the voiced velar plosive inserts U+0261 LATIN SMALL LETTER SCRIPT G, the letter the IPA uses, while your keyboard types U+0067 LATIN SMALL LETTER G. Many fonts draw them alike, but they are different characters, so a text search for one will not find the other. The X-SAMPA code g gives U+0261.
Why are code points and characters counted separately?: A symbol with a diacritic is one character on screen but two or more code points, and a character limit may count either. The readout shows both, plus the size in UTF-8 bytes: /ʃɪp/ /vɛ̃/ is 11 code points, 10 characters and 15 bytes.
Is my text uploaded or saved?: Nothing is uploaded: the page works on your text in the browser. Your draft and the tab you had open are kept in this browser's local storage so they are there next time, and the Start over button forgets them.

Character entry only: each key inserts the one Unicode code point it names, labelled with its Unicode 17.0 name, and your text copies as plain text and saves as a UTF-8 .txt file. It does not transcribe speech or decide how any word is pronounced. A combining mark such as U+0303 attaches to the symbol just before the cursor, shortcut codes follow X-SAMPA, and how a symbol looks depends on the fonts where the text is shown.