Account

What new tool do you want?

Text & Writing

IPA Phonetic Transcriber (English to IPA)

Convert English text to the International Phonetic Alphabet (IPA) in General American pronunciation, with stress marks — using the 135,000-word CMU Pronouncing Dictionary. Shows alternative pronunciations and ARPAbet.

How to use

  1. 1Type or paste English words or sentences.
  2. 2See the IPA transcription; hover a word for alternatives.
  3. 3Copy the result for lessons, dictionaries or linguistics work.

Frequently asked questions

Which accent does it use?

General American English, from the CMU Pronouncing Dictionary. British (Received Pronunciation) differs — for example, no 'r' after vowels (car /kɑː/) and a different 'o' in 'lot' (/ɒ/).

What about words not in the dictionary?

Names, new words and typos may be missing; they're shown in brackets so you can check them manually.

Why do some words have two pronunciations?

Words like 'read', 'live' or 'either' are pronounced differently by meaning or speaker. The first variant is shown, and alternatives are listed.

Sources

A worked example of why 'read' shows two entries

The CMU dictionary lists two separate pronunciations for "read" — R EH1 D (rhymes with "bed," the past tense) and R IY1 D (rhymes with "bead," the present tense) — because the dictionary is built from recorded speech and isn't aware of grammar or sentence context, so it can't tell which one your specific sentence needs. The tool shows the first listed variant as the primary transcription and lists the rest as alternatives in a separate column, which is a deliberate, honest way of surfacing genuine pronunciation ambiguity rather than silently guessing one and hiding the other possibility from you.

This is a 135,000-word dictionary lookup, not a letter-to-sound algorithm

Every transcription here is a lookup against the CMU Pronouncing Dictionary — a fixed list of real recorded American English pronunciations — loaded once as a roughly 1MB static data file from this site itself, not sent anywhere or generated by a model guessing at spelling-to-sound rules. That's precisely why a name, a brand, a typo, or genuinely new slang shows up in plain [brackets] rather than a best-effort guess: the tool only ever returns a transcription it found in the actual dictionary, and makes no attempt to algorithmically predict pronunciation for words that aren't in it.

Why British English speakers will see a different accent than they expect

The entire dictionary represents General American pronunciation specifically, which diverges from British Received Pronunciation in systematic, predictable ways — American English pronounces the 'r' after vowels (car as /kɑr/) where RP drops it (/kɑː/), and the vowel in "lot" differs between the two accents entirely. This isn't a limitation that affects a few edge-case words; it's a wholesale difference in the sound system, so a British English teacher or learner using this tool needs to consciously translate the output, not just double-check unusual-looking entries.

What the stress marks actually tell you, and why they move

A primary stress mark (ˈ) placed before a syllable indicates that syllable carries the main emphasis in the word, and a secondary stress mark (ˌ) marks a lighter, secondary emphasis — these aren't decorative, since English word stress genuinely changes meaning: 'record' stressed on the first syllable (ˈrɛkərd) is the noun, while stress on the second (rɪˈkɔrd) is the verb, and the CMU dictionary stores these as distinct entries precisely because the sound actually differs, not just the spelling.

ARPAbet exists because IPA symbols aren't plain ASCII

ARPAbet represents each phoneme using only standard keyboard characters (AA, IY, R, D and so on) specifically because it predates widespread Unicode support and was built for systems that couldn't render IPA's special characters at all — which is exactly why it's still the native format of the CMU dictionary and why this tool shows it as an alternative view alongside the IPA transcription. Anyone working with older speech-synthesis or speech-recognition tooling, much of which still expects ARPAbet-formatted input rather than IPA, needs that raw ARPAbet string rather than the prettier phonetic symbols.

Related tools

More: Text & Writing