Comparison · Updated August 2026
Kotoba vs Anki, honestly.
Same memory science, different bodies. Anki is the infinitely flexible screen SRS — best for kanji, custom decks, and anything visual. Kotoba is a purpose-built audio SRS — the same FSRS scheduling, run entirely by voice, for the hours Anki physically can't reach. Most learners who try both keep both: Anki at the desk, Kotoba in the car.
Disclosure: written by the developer of Kotoba. We use Anki too.
What they share
Both are spaced-repetition systems built on active recall, and both schedule with FSRS — each card or word comes back right before you'd forget it, based on your actual history with it. Both are free. If you believe in SRS, you don't have to choose a philosophy here; you're choosing an input device.
Where Anki wins
- Anything visual. Kanji recognition, kana, images, screenshots — the whole point of a screen. Kotoba can't do this and doesn't try.
- Custom decks. Your mining deck, your textbook's word list, medical terminology, whatever — Anki's card model handles it all. Kotoba ships curated JLPT decks only.
- Any subject, any language. Anki is a general-purpose memory tool; Kotoba is Japanese vocabulary, full stop.
- Two decades of ecosystem. Add-ons, shared decks, stats, sync across everything.
Where Kotoba wins
- The hours Anki can't touch. Anki needs eyes and thumbs for every card; it has no audio-only or background mode (the long version). Kotoba runs a full review session while you drive, cook, or walk — that's one to two otherwise-lost hours a day for most people.
- Real grading without touching anything. Speech recognition checks your spoken answer, so the scheduler gets honest data hands-free. Anki's auto-advance workaround grades every card the same, which quietly corrupts the intervals.
- Listening and pronunciation come free. Every rep is spoken and heard — native audio in, your voice out. A text card can't train your ear or mouth.
- Zero setup. No deck shopping, no note types, no add-ons: the JLPT lists with 40,000+ native clips and example sentences are the default. Open the site, tap Start.
Side by side
| Anki | Kotoba | |
|---|---|---|
| Interface | Screen — read, flip, tap | Voice — hear, speak, graded by ear |
| Scheduling | FSRS (or SM-2) | FSRS |
| Hands-free use | No (no audio-only/background mode) | Yes — designed for it |
| Content | Anything you build or download | Curated JLPT N5–N3, native audio + sentences |
| Kanji | Excellent | Not covered |
| Listening/speaking practice | Only if you build it in | Inherent to every rep |
| Setup effort | Real (decks, note types, options) | None |
| Platform | Desktop, AnkiDroid free, AnkiMobile $25 | Browser/PWA, free |
The split that actually works
Treat it as a schedule question, not a loyalty question. Desk time → Anki (kanji, reading, your custom material). Eyes-busy time → Kotoba (spoken JLPT vocabulary, listening, pronunciation). The overlap is small — a word you learn by ear in the car is easier to attach a kanji to at the desk, not redundant with it. If you only have desk time, stay with Anki. If your only free hour is the commute, Kotoba is the one that fits it — see how the audio loop works.
Keep your Anki habit — add the commute · No account needed
Frequently asked questions
Can I import my Anki deck?
Not currently — the audio-first session depends on curated native-recorded audio and sentences, which arbitrary decks don't carry. If custom decks are the priority, Anki (plus an audio-export tool like Audio Flash) fits better.
Will my Kotoba progress sync with Anki's scheduler?
No — they're separate systems with separate histories. In practice this doesn't hurt: the skills they train (spoken recall vs written recognition) are different enough that double-tracking a word is cheap.
Is Kotoba's grading as reliable as tapping Good/Again?
Speech recognition is right the vast majority of the time; when it's unsure it falls back to voice self-grading, which is exactly Anki's honor system — by ear. The scheduler data stays honest either way.