Kiku HoursCount the hours you understood.
Not the hours a video was open. Kiku Hours listens for the language you're learning and counts only that, on YouTube, Twitch and six more sites.
69 languages · 8 sites · free, no account · Google Chrome
Say what you're learning. Press play.
Choose your languages
The one you're learning, and the ones you already speak.
Watch like normal
Playback is counted from the player. Ads and pauses aren't.
It listens
Each video is labelled with its language, and only your target language counts.
Native-language time is recorded, not counted.
An English video on a Japanese channel goes in the log, but not the total. Nothing is deleted; the held-back time is shown next to the number it came out of, so the parts always add up.
Watch it add up.
A running climb for the year, every day on a calendar, streaks, and achievements dated the day you crossed them. Every figure comes from the same ledger, so no two numbers can disagree.
Every minute is yours, and stays on your machine.
Each interval is stored individually in your browser. Back up to a file and restore it on another computer, or export a spreadsheet. There is no account and nothing to pay.
Eight sites, one ledger.
Crunchyroll, Prime Video and Disney+ are next. See the roadmap.
Straight answers.
Is it really free?
Yes. There is no account, no subscription and no paid tier today.
Why Chrome only?
Kiku Hours is built for Google Chrome. In other browsers it still counts time; it just can't label the language yet. Other Chromium browsers are on the roadmap.
Where does the audio go?
Short windows of the video's audio are analysed to decide the language, and discarded. Nothing is recorded, and Kiku Hours never stores audio. Only the language label is kept, on your machine. Details in the privacy policy.
Which languages?
Sixty-nine, including Japanese, Korean, Mandarin, Cantonese, Spanish, French, German, Portuguese, Italian, Russian, Arabic, Hindi and Thai. Each was measured against real speech before the list was set.
How accurate is the language detection?
Conservative by design. A wrong label that lands on your native language would remove hours, so nothing is believed until two independent checks agree. When it isn't sure, it counts the time and leaves the label off rather than guessing.
What about videos with loud music?
A voice buried under loud music can't be identified. On YouTube, Kiku Hours falls back to the language of the video's auto-generated captions after a couple of minutes; elsewhere the time is counted without a label.