From OpenStreetMap Wiki
Jump to navigation Jump to search
Public-images-osm logo.svg name:pronunciation
Bourg-en-Bresse phonétique.jpg
Phonetic transcription of a name into the International Phonetic Alphabet (IPA). Show/edit corresponding data item.
Group: names
Used on these elements
may be used on nodesmay be used on waysmay be used on areas (and multipolygon relations)may be used on relations
See also
Status: in use

This tag contains a phonetic guide to pronouncing the name contained in name=*. The name is transcribed into the International Phonetic Alphabet (IPA), either using a broad transcription or a narrow transcription.


This tag can be useful for clarifying an idiosyncratic pronunciation. For example:

  • Most places named Reading are pronounced like "redding" in English, not like "reeding". This pronunciation extends beyond places to the roads and points of interest named after them.
  • Houston Street in New York City is pronounced very differently than the city of Houston, Texas, or the Scottish village of Houston, Renfrewshire.
  • The Vietnamese given names Dũng and Loan have very different pronunciations than the English words "dung" and "loan" but often appear in English-language street and POI names in some parts of the United States. way 500813798 node 7164922833
  • Some business names contain unit symbols that look similar to words. (In general, name=* expands abbreviations, but nevertheless abbreviations in business names and other trademarks are commonly left alone.) node 7979223615

End users may benefit from seeing a pronunciation hint below a place’s label on a map, or from hearing the place name correctly read aloud while navigating with a routing application. Blind users also rely on software to read map features aloud. A renderer or geocoder can use an unambiguous phonetic transcription to automatically transliterate from an irregular orthography like English to another orthography. A geocoder can also use the phonetic transcription to generate "did you mean" results according to a phonetic algorithm.

This tag was first used in 2008 on various English place POIs.


As with name=*, an ISO 639 language code can be included in the key to specify the pronunciation in a particular language. For example, if a feature is tagged with name:en=* and name:fr=*, name:en:pronunciation=* and name:fr:pronunciation=* can indicate the pronunciation in English and French, respectively. The :pronunciation suffix is also used on a several other name-related keys, such as official_name:pronunciation=* and destination:pronunciation=*.

name:iso15919=* is another pronunciation key in use, for the ISO 15919 transliteration standard for Romanized transcription of Indic/South Asian languages in particular. name:kn:iso15919=* is the most prevalent related key in use at the time of writing, for transcriptions of Kannada language names. The ISO 15919 system may be better suited for indicating pronunciation for Indic languages as it includes provisions specific to the phonetic elements they have in common. Bear in mind that while various letter-to-letter transcription charts exist, these serve as guides rather than exact mappings. If we were to use just the characters to derive a transcription of the same name in Urdu and Hindi, we would get a different outcome due to differents in the way the source scripts encode phonetic information.

As an alternative to tagging a feature with a name:pronunciation=* tag, the wikidata=* tag allows you to associate a feature with the Wikidata item describing the same feature. That item can be tagged with an IPA transcription statement. However, note that Wikidata items are subject to different inclusion criteria than OpenStreetMap features.


Software support

OSRM v5.2 and above includes the value of this tag in the RouteStep object. Valhalla exposes the pronunciation in a similar manner. [1] Mapbox Directions for Swift exposes this value as RouteStep.phoneticNames when connecting to an OSRM-powered server.

Clients of these libraries can pass the transcription into any text-to-speech engine that supports IPA, which is the default phonetic alphabet of the Speech Synthesis Markup Language (SSML). [2] A number of text-to-speech engines support IPA transcriptions, including:

The open source eSpeak and MaryTTS projects both support Kirshenbaum, an ASCII representation of IPA.

The Mapbox Navigation SDK for Android and iOS applies IPA transcriptions from name:pronunciation=* when announcing turn instructions using Amazon Polly. The iOS version also applies IPA transcriptions when falling back to AVFoundation when Polly is unavailable. [3]

Editors and data consumers may need to special-case this key and other subkeys of name=* that do not represent localized versions of the name=* key. [4]

Possible synonyms

As recorded in the IANA Language Subtag Registry, IETF BCP 47 encodes IPA pronunciations as <language>-fonipa, which would imply subkeys such as name:en-fonipa=*. As of March 2021, such subkeys have never been used and are unsupported in software. BCP 47 requires an explicit language code (primary language subtag), whereas name=* does not explicitly indicate a language. If multiple *:*-fonipa=* tags are specified, data consumers would find it more difficult to associate one of them with name=* unless *:und-fonipa=* is also specified.

See also

External links