Titles & names

The original title is primary

Every listing shows a work under its own name in its own script — Japanese for anime, Chinese for donghua, Korean for aeni — with the best Latin-script name beside it. A romanisation is a rendering for readers who cannot read the original; it is not the work's name.

Script of the primary title

han means Han characters with no kana and no hangul. That distinction carries weight: Japanese kanji and Chinese hanzi share a Unicode block, so an all-kanji Japanese title is genuinely ambiguous read alone. Chinese is therefore only ever inferred for a string sitting beside a title already known to be Japanese.

Chinese titles

7,853 of 16,010 titles carry one (49.1%). No anime API exposes a Chinese title, so this is assembled from outside sources in a fixed order of authority — a stronger source always overrides a weaker one, and the winner is recorded on every record.

How accurate is it?

Measured, not asserted. A stratified sample of 312 mappings was audited title by title, and every flagged case went to an independent reviewer instructed to defend the pipeline, so one reviewer's false positive could not become a reported defect. 22 were flagged; 17 survived.

The dominant failure was not a wrong translation but a wrong language: Japanese titles written entirely in kanji. A kana check cannot catch them, so the filter inspects characters — roughly 130 Japanese-only shinjitai (総 is neither 总 nor 總), strings mixing simplified-only with traditional-only characters, and Latin-dominant values are all rejected. Removing the weakest source cost about a thousand titles of coverage: the intended trade, since a missing title is a gap you can see and a wrong one is not.

Where the coverage is

Coverage tracks how much anyone cared to translate a title. The well-known catalogue is nearly complete; music videos are essentially untranslated everywhere.

Simplified or traditional

Classified by character set — an 827-pair table of characters existing in only one script. Where a title uses only shared characters the text cannot say, and the source decides instead; anything still undetermined stays unknown rather than guessed.

Every name, side by side