Pronunciation Data Sources and Licenses
Sources in this build
General American (data version lx1-850f78dd69af)
- ipa-dict: Monolingual wordlists with pronunciation information in IPA (en_US)
- ID:
ipa-dict - Pronunciation candidates: the en_US word list with IPA.
- Snapshot:
43c3570eb3553bdd19fccd2bd0091534889af023 - Licence:
MIT. Reviewed and approved on 2026-09-27. - Attribution: Pronunciations from ipa-dict (open-dict-data), en_US, based on cmudict-ipa by lingz and the CMU Pronouncing Dictionary (Carnegie Mellon University).
- Project page
- cmudict-ipa (lingz): CMUdict converted to IPA, upstream of ipa-dict en_US
- ID:
cmudict-ipa - The IPA conversion of the CMU Pronouncing Dictionary from which ipa-dict en_US was made (attribution chain).
- Snapshot:
629483e9a7fa5b9ff11bb83e3fda6dcc25c5c511 - Licence:
MIT. Reviewed and approved on 2026-09-27. - Attribution: cmudict-ipa by lingz: IPA conversion of the CMU Pronouncing Dictionary.
- Project page
- CMU Pronouncing Dictionary
- ID:
cmudict - Origin of the ipa-dict en_US pronunciations; used to cross-check coverage while building the data.
- Snapshot:
74790861f652b15e4ac49015a90074ad62a27690 - Licence:
LicenseRef-CMUdict. Reviewed and approved on 2026-09-27. - Attribution: The CMU Pronouncing Dictionary, Copyright (C) 1993-2015 Carnegie Mellon University.
- Project page
British RP-style: not available in this build, so no source is listed for it.
Font
- Charis 7.000 (SIL International)
- The font used to display IPA.
- Licence:
OFL-1.1. Reviewed and approved on 2026-09-27. - Attribution: Charis 7.000 by SIL International, under the SIL Open Font License 1.1.
- Project page
Licences, attribution and downloads
Your own text is yours. It is not part of the pronunciation data, it is not sent to this site and it is not licensed to anyone by using the tool. The pronunciation data comes from the third-party sources above, each under its own licence; they are not all under one licence.
Changes made to the data: the symbol conventions listed on the methodology page are applied, pronunciations outside the notation are held back, and the entries are stored in a compact file per system. Every transcript you copy or download names the sources it used.
- General American: licence texts, data manifest and pronunciation data (125,927 words).
Corrections
Your edits in a transcript stay in that transcript and in the files you save. They never change the public data, and words that are not in the data are not collected.
To suggest a correction, use Help → Send feedback. You choose the category and whether to include the word, a short context and a message; nothing else is added. A suggestion is checked by a reviewer before any change, a disputed entry needs two reviewers, and an accepted change reaches the data only in a later data version listed in the changelog.