About
Hi, I'm Ed! OnJapanese is a shadowing app that I made for myself. Here is how it works: you listen to a Japanese phrase, say it back, and then hear the two played one after the other (first the model voice, then you). That loop — listen, speak, compare — is the whole idea, repeated on a small set of phrases over days until they come out smoothly.
I have a lot more I want to build on this site, so this is just a start (it also means some things might be broken).
The voices
Every phrase is spoken by a synthetic voice — not a recording of a person — made with VOICEVOX, a free, open-source Japanese speech engine. I've given each one a short name so they're easy to tell apart; the real VOICEVOX name is shown below it.
-
Naomi
VOICEVOX Nemo · 女声1 -
Hiro
VOICEVOX:雀松朱司 -
Rena
VOICEVOX:波音リツ -
Makoto
VOICEVOX:玄野武宏
Each of these voices is used in accordance with its own terms of use.
Your voice stays on your device
When you record yourself, the recording is held in your browser's memory just long enough to play it back to you. It is never uploaded, never saved to a server, and it disappears when you move to the next phrase or close the page. There are no accounts, and nothing to sign in to.
Which phrases you've practiced, where you left off, and which voice you prefer are remembered on your device only — so they'll be here tomorrow on the same browser, and won't follow you to another (unless you clear your browsing data).
Where the content comes from
The grammar-pattern names are compiled from several community-maintained JLPT lists. The explanations and every example phrase are generated fresh for this app rather than copied from those sources, and I review them by hand before the audio is made — but I'm not a native Japanese speaker. Mistakes are mine; if a phrase reads oddly, it probably needs another pass.