Speak English every day, and prove it.
Wake forces a daily conversation, picks the topics for you, and measures what actually improves while a voice model holds the other end.
Your choice of topic.Five seconds of silence.A good accent can hide a hole
There's a profile classic methods serve badly: someone with the accent, rhythm and syntax of a native speaker, whose vocabulary has narrowed to the contexts they actually use.
They talk about strategy, politics or projects effortlessly — then stall on colander, hem, skirting board, direct debit, sprain.
That isn't a level problem. It's a coverage gap, and lessons don't treat it: they teach grammar progression to someone who doesn't need it.
- TypeInstallable web app
- StorageOn the device
- AccountNone
- Target varietyBritish English
- Daily ceiling13 min of speech
- StatusVoice selection under way
Six principles, and no lessons
The partner never corrects head-on and never explains a rule of grammar.
- 01The topic is assigned
Never chosen: choosing always leads back to the comfortable vocabulary you already own.
- 02No head-on correction
The partner reuses the right form in its own reply and moves on. Three times a session at most.
- 03Help is earned
Five to seven seconds of silence on a single word — long enough to find it or work around it. Immediate help on an idiom: those can't be deduced.
- 04Three ledgers
A word counts as acquired on its third spontaneous use, an expression on its fourth across four situations, a grammar pattern at eight occurrences out of ten.
- 05Independent measurement
A separate layer reads the transcript and computes the avoidance rate. A conversation partner is structurally forgiving; it can't be its own judge.
- 06An asymmetric lock
Raising the difficulty takes effect immediately. Lowering it takes effect seven days later. You don't drift downward.
Hard ceilings, everywhere
The session cuts off at the planned length even when the conversation is going well. What protects the habit is exactly what stops you overdoing it on the good days.
Three weeks pass with no measurement and no feedback at all. The only goal is coming back the next day; the baseline is set on day twenty-one.
The curve goes up before it goes down
A control topic is drawn on day twenty-one, kept out of rotation, and revisited monthly. It's the only comparison in the system that can't be argued with.
A warning appears at that point, and it matters: once you stop avoiding the words you're missing, you meet more of them. The avoidance rate rises before it falls. Without the warning, people quit at precisely the moment the thing starts working.
Easiest thing is to try one
The games run in your browser, free. The apps are in the catalogue.