Hi everyone,
I’ve been building a free Japanese Ebook reader/furigana converter, and I’d like to see if WaniKani users can break its furigana 
One problem that kept bothering me with automatic furigana is ambiguous readings.
For example:
大分に旅行した。 → 大分【おおいた】
大分よくなった。 → 大分【だいぶ】
Or:
市場で魚を買った。 → 市場【いちば】
市場規模が拡大した。 → 市場【しじょう】
A dictionary knows that both readings exist, but it does not necessarily know which one belongs in the sentence.
EZFurigana tries to choose the reading from the surrounding context instead.
You can try it here:
What it does
Right now you can:
-
paste Japanese text and generate furigana
-
choose various file formats (PDF, DOCX, EPUB, PNG)
-
adjust how furigana is displayed with JLPT filtering
-
click words for dictionary information
-
export and download files
The main thing I’m trying to achieve is furigana you can actually trust while reading native material, rather than simply attaching the first dictionary reading to a word.
It definitely still makes mistakes. Names, unusual compounds, and rendaku are especially difficult.
Want to help me test it?
The most useful thing you could do is very simple:
Try to find a sentence that makes it choose the wrong reading.
If you find one, paste the sentence here along with what the reading should have been or use the in-built reporting feature. Those examples are extremely useful for improving the system.
I’m also curious about one WaniKani-specific idea:
Would a mode that hides furigana for kanji or vocabulary you’ve already learned in WaniKani be useful?
I know tools like Yomitan already solve parts of the native-reading workflow extremely well, so I’m not trying to replace them. My goal is mainly to make processing Japanese files and reading them a little easier.
Any criticism is welcome, especially sentences where it gets the Japanese wrong.