Hear the difference.
Find your voice.
?
Practice a short passage, listen to a reference, record your reading, and compare. No ebook required.
Read an EPUB book
?
Choose a text-based, unencrypted EPUB from Files. The book stays on this device. Select up to 500 characters and tap Use selection for practice. Only explicit speech or translation actions send selected text to a service.
?
Small hiragana readings above kanji are generated on this device. Names and context may differ. Your choice is saved.
?
When pasting an Apple Books excerpt, removes its outer quotation marks and book/copyright footer. Your choice is saved on this device.
edit this or use a sample
?
Use all passage text instead of one sentence. Google playback uses short clips in sequence; downloaded Google comparison audio still has a 200-unit limit. Recordings and examples for the whole passage are saved separately.
?
Choose one sentence below. Each recording is up to 20 seconds.
?
Keep the current text and visible Japanese readings at the top while scrolling. Long content scrolls inside this panel. Your choice is saved on this device.
?
Tap Translate to send the current sentence or whole passage to Google. Choose a target language and copy the result. Showing this panel alone sends nothing. Experimental service; automatic translations may be inaccurate.
?
Automatic dictionary readings may differ for names or context. Readings are generated on this device.
?
English and Japanese. Google speech is selected when the page opens; browser voices are also available.
Play reference or Add Google comparison audio sends this sentence to Google. Downloaded clips are saved on this device for comparison; recordings are not uploaded. Experimental, unofficial service; availability is not guaranteed. Maximum 200 UTF-16 units per sentence. Shorten longer sentences or use browser speech.
?
Listen to the current sentence with the selected provider.
Playback diagnostics · troubleshoot audio
?
Open isolated native audio check · compare Google speech with a same-site tone.
Local-only log of the latest 120 events. No passage text, audio, filenames, full URLs, or raw error messages. Nothing is sent automatically. Clear or refresh to discard. Copy only if you want to share it.
?
Sends the current sentence and language through this site's audio service to the configured backend. The backend chooses the AI voice. Saves the returned audio on this device for playback and comparison.
?
Add a teacher or native speaker clip for the selected sentence.
?
Direct audio capture keeps other audio and video playing. For shared audio, choose a source and enable Share audio if offered. Mac app audio in Firefox uses the local helper: run npm run capture:setup, then npm start on this Mac. macOS asks for Screen & System Audio Recording permission. Captures up to 20 seconds of system sound; no screen frames or microphone are saved.
?
Tap to record. Tap again to finish.
?
Latest three plus one kept · saved on this device
?
Choose two real audio clips of the same sentence. See their full waveforms on one time scale, listen, and compare their timing and energy.
All analysis stays on this device. Temporary imports are not uploaded or saved. Each clip must be 60 seconds or shorter and at most 10 MB. Add Google comparison audio above to compare its speech clip, or import or record other audio. Browser speech remains playback-only.
Choose two saved clips or import local audio to compare.
?
optional, saved locally