Open a voice memo on your phone and record yourself speaking for one minute about your day. This baseline recording gives you a starting point to compare against, so later progress becomes visible instead of a feeling you have to guess at.
The practice loop
Record one minute on any topic without stopping. Do this once to warm up, since starting cold makes the first attempt unrepresentative of your real fluency.
Speak on the same topic for four minutes. Repeat it for three, then two. The shrinking time pressure pushes faster retrieval, and studies with second-language learners found fluency gains over five to seven weeks when the same speech was repeated.
Repeat steps 1 and 2 three times across the week, keeping the same topic within each session. Change topics between weeks.
Relisten to two specific recordings: the warm-up minute from step 1 today, and the last minute of your two-minute pass. The same ideas should sound smoother in the second one. Hearing the same speech at two speeds gives you immediate, concrete feedback, which deliberate practice research treated as central rather than relying on total hours spent.
If four minutes feels out of reach at first, halve every interval and keep the ratio. There is more than one valid way to do this drill, and the published gains come from small studies, so treat your own recordings as the main evidence that it works for you.
A complete run-through
Pick one small topic, like what I ate yesterday and why. Small topics work because you never stall searching for ideas; the difficulty stays on speaking, not thinking.
First, record one warm-up minute on anything. Then start the timed drill. Speak the same topic for four minutes without stopping, even if you repeat yourself. Pausing to fix mistakes is fine, but keep the recording running.
That first pass can feel awkward and mistake-prone. Research on repeated speech suggests that feeling tends to fade across repeats, and motivation often rises as mistakes lose their sting, so keep the timer running.
Repeat it for three minutes. You will notice you cover the same ground faster, because the four-minute version you just recorded is already in your head. If you finish early, keep talking until the timer ends.
Repeat it once more for two minutes. By now you should cover the whole topic with fewer long pauses and less searching for words. That speed-up across passes is the entire point, so let it happen rather than adding new content.
A full pass looks like this. In the four-minute recording you say, I usually skip breakfast, but yesterday I made eggs because my meeting was late. In the two-minute version, the same idea comes out earlier, shorter, and with fewer restarts like um, I mean. It might sound like I usually skip breakfast, but yesterday I made eggs because my meeting was late, delivered in one smooth stretch with no false starts. Because published gains come from small studies, treat your own recordings as the main evidence it works for you. Then listen back to the two-minute recording. Count two things: the longest pause you hear, and one phrase that came out more smoothly than in the four-minute version. Write both down in one line, for example, pause at 0:40, breakfast phrase came out clean.
Self-check and next adjustment
Compare your one-line notes across sessions. If the longest pause gets shorter each time, like 0:40 this week versus 1:10 last week, your retrieval is speeding up and the routine is working for you.
A second check: replay the two-minute recording next to the warm-up minute from the same day. The same ideas should sound smoother and start sooner. If you hear it, trust that evidence over any feeling of progress.
One adjustment, depending on what you hear. If pauses stay the same length, shrink the interval to 3/2/1 instead of 4/3/2. Same ratio, less time per pass, so the pressure to speak faster returns without demanding four straight minutes.
If repeating the same speech starts to feel tedious and your pauses grow instead of shrink, do not abandon the drill. Instead, plan a different topic for next week's sessions, so each week you repeat one speech several times within the week and then start the following week with fresh material. Keep one topic per session and change topics only between weeks, because the repetition of the same speech is what drives the fluency gains in the studies. Boredom was one reported reason a similar study found no improvement, so watch for that sign and refresh the material week by week. The principle stays the same: keep the repeat, change the material.
Whatever you adjust, change only one thing per week and keep recording. With one change at a time, the next week's pause count tells you clearly whether the adjustment helped or not.