Sources and licence

Dictionary data

SourceWiktionary, the English-language edition, Urdu sections
Extractionkaikki.org Urdu dictionary, produced with the open-source wiktextract tool (file last modified Fri, 30 Oct 2026 23:14:51 GMT, downloaded 2026-11-22)
LicenceCreative Commons Attribution-ShareAlike 4.0
AuthorsWiktionary contributors. Each entry page links to its Wiktionary page and its history, which lists the authors.
Size10,507 raw entries; 7,596 Urdu-script headwords after merging parts of speech; 2,944 with a full page

What Lughat changed

  • Merged entries for the same headword and removed non-Urdu headwords (for example Judeo-Urdu in Hebrew script).
  • Cleaned etymology text: removed "etymology tree" dumps and the glyphs of historical scripts (Brahmi, Old Persian, Devanagari and others) that no shipped font can draw, keeping the transliterations.
  • Derived a "Roman Urdu as typed" spelling from Wiktionary's scholarly transliteration by removing marks and mapping letters such as ś to sh. This is a mechanical rule, not part of Wiktionary.
  • Chose which entries get a full page: a meaning plus at least two of an etymology, an example, synonyms, word forms, related words or several senses.

Because Lughat's entries are an adaptation of Wiktionary content, the entries on this site are also shared under CC BY-SA 4.0: you may copy and adapt them if you credit Wiktionary and share your adaptation under the same licence.

Original content

The guides, hub pages, alphabet notes and the converter are written by DanixSoft. The converter's core word list (very common function words such as hai, nahi, kya) was checked by hand. The code is not under the CC licence.

Pronunciation audio

Where an entry has a recording, it is a Lingua Libre recording hosted on Wikimedia Commons. It is loaded only when you press the play button, directly from Wikimedia, and its licence and speaker are on the Commons file page.

Converter accuracy (measured)

We measure the converter on the dictionary itself. Half of the headwords are removed from the dictionary at random (by a fixed hash), the converter is asked to convert their Roman spelling, and the result is compared with the real Urdu spelling, ignoring vowel marks.

TestWordsExact matches
Words found in the dictionary (all headwords)6,28896.2%
Spelling rules only, careful typing (long vowels doubled)2,10155.2%
Spelling rules only, casual typing (no long vowels marked)3,16023.5%

The words in the "rules only" rows are rarer words than average, so real-life text, which is dominated by common words the dictionary knows, converts better than these numbers suggest. A mismatch is not always wrong (several spellings can be acceptable) but we count it as a miss.

Not used

Lughat does not copy text from Rekhta, Oxford, Platts or other dictionaries, and it uses no machine-generated meanings.