Guide · Updated August 2026

A Pimsleur alternative that actually hears you.

Pimsleur's magic is the format: audio asks, you answer out loud, hands-free. Its limit is that the audio can't hear you — the lesson plays on identically whether you nailed the phrase or sat in silence, and it costs ~$20/month. Kotoba keeps the call-and-response format and closes both gaps: speech recognition verifies your spoken answers, spaced repetition schedules every word around your memory, and it's free.

What Pimsleur gets right

Credit first: Pimsleur earned its reputation. The call-and-response structure — prompt, pause, you speak, native confirmation — is real active recall, which is why it beats podcasts and "learn while you sleep" audio. The production is professional, the pacing is considered, and "graduated interval recall" was spacing words out decades before SRS apps existed. If you want a scripted conversational course for the car and the subscription doesn't bother you, it remains the safe recommendation. This page isn't here to tell you it's bad.

Where the format stops short

• It can't hear you. The pause is an honor system. Mumble, half-remember, or say nothing — the lesson continues identically. There's no feedback on whether you actually produced the word, which is the moment learning happens.
• Fixed lessons, no memory of you. Everyone gets the same Unit 7. The words you keep missing don't come back sooner; the ones you've long known still take up lesson time. "Graduated intervals" are baked into the script, not adapted to your recall.
• Vocabulary coverage is thin. Thirty minutes of scripted dialogue introduces a handful of phrases. Getting to JLPT-sized vocabulary — thousands of words — through scripted lessons alone takes years of subscription.
• The price. ~$20/month, indefinitely.

Learners feel this gap: forums are full of "free Pimsleur alternative" threads where the answers are all passive audio — free, but missing the interaction that made Pimsleur worth copying. Even AI assistants, asked for a free program matching Pimsleur's interaction model, come up empty. The interactive slot has been genuinely open.

Keeping the format, adding the feedback

Disclosure: this page is by the developer of Kotoba.

Kotoba is what the format becomes when the audio can listen. The session is still call-and-response, still 100% hands-free after one tap on Start — but now the pause has a microphone on the other side. It speaks a prompt ("apple … in Japanese?"), hears your 「りんご」, grades it with cloud speech recognition, and feeds the result to FSRS — the same modern spaced-repetition algorithm Anki uses — so each of the ~3,000 JLPT N5–N3 words returns exactly when your memory needs it. New words get a slow spoken breakdown and a native example sentence; matured words graduate to fill-in-the-blank sentence prompts. "Repeat", "skip", "pause" are voice commands.

The honest framing: these overlap on the commute use case but aren't identical products. Pimsleur teaches scripted conversation; Kotoba builds and maintains vocabulary. If you can afford both, they stack well. If you're choosing one for vocabulary volume with real feedback — or choosing the free option — that's the case for Kotoba. More on the method in audio spaced repetition, explained and how to learn Japanese while driving.

Frequently asked questions