The drill that works with nobody in the room is to say the same thing three times, giving yourself less time on each pass. Four minutes on one subject, then the same material in three, then the same again in two. The content is held constant so that the only variable left is how fast you can get at it.
That constraint is what makes it work. Any exercise where both the subject and the time change is measuring nothing, because you cannot tell an easier topic from a faster retrieval. Repeat the identical content and the improvement between the first and third pass is the thing you came for, and you can hear it.
Eight ways to practise with nobody there
- narrate what you are doing, out loud, while cooking or walking. It is the cheapest speaking practice available and it builds the habit of not translating.
- read aloud for five minutes daily, which trains the mouth even though it trains no other part of speaking.
- record a voice note daily on one topic, and keep them, because the month to month comparison is the motivation.
- argue both sides out loud. Taking a position and then the opposite forces vocabulary you would otherwise avoid.
- describe a photograph for sixty seconds, which is also the exact task several speaking exams set.
- think in English at a fixed time, on a commute for example, so it becomes a habit rather than an intention.
- use a mirror for stress, not for looks. Watching your own jaw shows whether you are producing the vowels or approximating them.
- talk to a recording and answer it, pausing a podcast to reply, which is the closest thing to a conversation you can do alone.
What is genuinely available alone
Three things. Retrieval speed, by the drill above. Rhythm, by borrowing somebody else's. And the phrase store, by saying the same fixed sequences enough times that they come out whole.
What is not available is anybody else's behaviour. You cannot practise the pressure of a turn that is about to end, or the repair of a misunderstanding you did not know was happening, or the experience of being interrupted. Those are the parts a room supplies and no drill substitutes for them, which is worth saying because the alternative is a learner who drills for a year and is still surprised by the first real conversation.
There is one thing alone practice does better than a partner, and it is not a consolation. You can stay on a subject you are bad at. A polite person rescues you within thirty seconds by changing the topic or finishing your sentence, and their kindness is why most conversation practice quietly avoids the hard subjects. Alone, nobody rescues you, so you can spend twenty minutes on how something at work is broken, which is exactly the material a real conversation will demand and no exchange ever gives you.
Where it stops: reading aloud
The stall is substituting reading for speaking, and it happens because reading aloud is pleasant and speaking into an empty room is not. A learner reads a page of English out loud, the sound is fluent, the pronunciation is being practised, and it feels like the same activity.
It is not the same activity in the one respect that matters. Reading supplies every word, so the retrieval step, which is the entire difficulty of speaking, has been removed. Somebody who reads aloud daily for a year will have better pronunciation and exactly the same speaking problem, and will not understand why.
The test is simple enough to apply to any drill you are considering: if the words are in front of you, it is not speaking practice. It may still be worth doing. It is not this.
The recorder nobody wants to play back
The voice recorder already on your phone is the whole toolkit, and the reason to name it is that it replaced the language lab's tape booth, which is where this drill used to have to happen and which almost nobody now has access to. Free, installed, and no setup.
Its limitation is not technical. A recorder records and does not judge, so the value only arrives when you listen back, which is uncomfortable enough that most people record diligently and never play anything. If you are going to skip the listening, skip the recording too and just do the drill; the tape that is never heard is a ritual.
What to listen for is narrow. Not errors, which you will hear plenty of and cannot fix in the moment anyway. Where the pauses fell, and whether the third pass was genuinely faster than the first.
What it looks like when it goes wrong
The recognisable failure is the four-minute pass that lasts ninety seconds because you ran out of things to say. That is a subject problem rather than a language problem, and the fix is to pick something with detail in it: a process you know, an argument you have had, how a thing at work actually gets decided.
The second failure is a third pass that is shorter because material was dropped rather than because delivery got faster. Cutting content is the easy way to hit the clock and it defeats the design, so the rule is that the third telling has to contain everything the first one did.
Borrowing somebody else's rhythm rather than generating your own is speaking along with a recording, and what it fixes, and what you are measuring on the playback is set out in speed without accuracy, accuracy without speed, and how they meet. Both sit under getting from correct sentences to an actual conversation.
The open question is the one this drill is designed around rather than answering. Everything above buys you speed on material you chose, in a room with no risk in it, and the reason most learners need that room at all is a separate problem with a separate mechanism: where the freeze comes from and what dissolves it.