Most flashcard sessions end with a tap. Colloquial ends each card with your voice, because when the card arrives as a phone call, the phone is already at your ear.
Every flashcard app I’ve used ends a card in roughly the same way. You reveal the answer, decide how well you did, and tap a button. Anki gives you four choices: Again, Hard, Good and Easy. Most apps give you two.
It feels normal, but that little tap makes a couple of assumptions: that you have a hand free, and that you’re looking at your screen at the exact moment you finish thinking. Neither holds on a phone call.
In Colloquial, a practice session arrives as an incoming call. Once you answer, the phone goes where phones usually go during calls: against your ear with the screen dark, or into your pocket while you walk with earphones in. That’s deliberate, because the whole point is to practise during time you were never going to spend staring at a screen anyway, and it meant the rating button had to disappear. What replaced it has become one of the parts of the app I’d least want to lose.
The call plays the question in the voice recorded for that card, and then it goes quiet and waits for you to answer out loud, as if someone had asked you.
The pause isn’t always the same length. A card with a longer answer gives you more time, and a card you’ve been getting right for weeks gives you less, so the easy ones can’t slide into autopilot.
After the pause the call plays the correct answer, and you hear how close you came. Sometimes the gap is tiny and sometimes it’s painfully obvious. You say “got it” or “needs work”, and the next card begins.
You never have to look down.
This feature came directly from the problem that made me build Colloquial in the first place.
My Cantonese flashcards were working, at least according to the app. I was learning more words and remembering more cards, but I was getting better at recognising written prompts rather than at understanding Cantonese, and when someone used those same words in conversation at a normal speed, I often missed them entirely. Reading a card and tapping a button had never asked me to recognise the word by sound before seeing it, or to produce it with my own mouth.
Answering out loud works on the speaking half of that problem, and hearing the question instead of reading it works on the listening half.
Voice rating is what holds the whole thing together. If every card ended with a glance and a tap, the session would keep pulling you back towards the screen, and after a few cards it would stop feeling like a call and start feeling like another app you happened to be using through your headphones. When you rate the card with your voice, the call remains a call from beginning to end. Your phone can stay locked, the screen can stay off, and the whole session can happen with the phone still at your ear.
Speech recognition brings up an obvious question about where your voice goes, and I wanted the answer to be simple.
Colloquial only enables voice rating on phones that can recognise the relevant phrases on the device itself, and it checks for that before it starts listening. Nothing you say during the call is sent away from your phone. The app uses your voice for two things, moving the small level meter on the screen and telling “got it” apart from “needs work”, and nothing else.
If your phone can’t perform on-device recognition for your language, you can use the two buttons instead, and everything else works in exactly the same way.
Getting a phone to recognise two short phrases may sound simple. It hasn’t been.
Doing it reliably in the middle of a call, across different manufacturers and different speech-recognition systems, has been the hardest engineering problem in the app. Harder, unexpectedly, than making the phone ring.
The two mobile platforms handle speech differently, some phones need an additional language download before they can recognise anything, and calls take control of the microphone in ways that ordinary apps rarely have to deal with. I’m still fixing problems on particular devices. When voice recognition works, it makes the session feel remarkably natural, and when it doesn’t, the buttons are always there.
If you use flashcards for something you’ll eventually need to say or hear, try answering every card out loud for a week before revealing the answer. It might be a language, medical terminology, lines for a role, or a passage you want to learn by heart. You don’t need Colloquial to try the idea, so use whichever app you already have.
It feels a little awkward at first, and gradually the cards stop feeling like things you can recognise on a screen and start feeling like things you can say.
Colloquial puts that entire loop into a phone call that rings at the time you choose. You can use your own recordings or one of the ready-made decks, answer each card out loud, rate it with your voice, and finish the session without looking at the screen.
Colloquial is a flashcard app that calls you and walks you through your deck out loud.
See how it works →