Last updated: 2026-07-26
Attribution
Mirasen's lessons are written in-house, but some of the reference data behind them comes from open Japanese-language projects. This page names every one of those projects, the licence it is used under, and the point where our own work begins.
Open Linguistic Data
Kanji readings, meanings and stroke counts in our reference data are derived from KANJIDIC2, published by the Electronic Dictionary Research and Development Group under the Creative Commons Attribution-ShareAlike 4.0 licence. Those fields remain available under that same licence.
Stroke-order diagrams are KanjiVG, copyright © 2009–2026 Ulrich Apel, used under the Creative Commons Attribution-ShareAlike 3.0 licence.
JMdict/EDICT is used while we build the app to check our vocabulary against a reference dictionary. No JMdict text is published inside Mirasen.
Pronunciation Audio
Every spoken line in Mirasen is synthetic speech. It is generated with Azure AI Speech on our own machine before a release and shipped as audio files, so the app never calls a speech service while you use it, and there is no human recording and no narrator behind any voice. These are the Japanese voices we use.
- Azure AI Speech: Nanami (ja-JP, neural HD and neural)
- Azure AI Speech: Aoi (ja-JP, neural)
- Azure AI Speech: Mayu (ja-JP, neural)
- Azure AI Speech: Shiori (ja-JP, neural)
- Azure AI Speech: Masaru (ja-JP, neural HD)
- Azure AI Speech: Keita (ja-JP, neural)
- Azure AI Speech: Daichi (ja-JP, neural)
- Azure AI Speech: Naoki (ja-JP, neural)
- Azure AI Speech: Masaru Multilingual (ja-JP, neural), the listening narrator
Written by Us
Every lesson, grammar explanation, example sentence, reading passage, exam question and mnemonic on Mirasen is our own work. The open data listed above stays with its original authors and is used under the licences named here.