Shadowing is the practice of repeating a speaker out loud, line by line, a breath behind the original. The whole method lives or dies on one rule: speak the next line aloud 1-2 seconds behind the original speaker, matching stress and intonation, instead of reading silently. Do that on every repetition and pronunciation starts moving within weeks; mouth the words, and the same minutes train almost nothing. This guide stays on that rule — how to do each repetition correctly, not how to schedule them.
If your bottleneck is the calendar rather than the technique, the ten-minute daily loop for busy learners covers scheduling; everything below assumes a clip is already open.
Pick a clip your ear can almost hold
Choosing material is a numbers decision. Take a 30-60 second clip that you understand about 90% of without subtitles: one speaker, clean audio, minimal background noise. The 90% threshold keeps the technique alive. Understand far less and your attention collapses into decoding, the delay breaks, and shadowing degrades into pausing after every word. Shadowing works through close repetition of a little audio, so thirty seconds looped gives depth, and depth is what moves pronunciation.
The four passes over one clip
- Choose the clip. 30-60 seconds, one speaker, about 90% understood without subtitles — hard enough to stretch you, easy enough to keep the 1-2 second delay alive.
- Pass the clip line by line. Hear the whole clip once without speaking, then play it again and speak each line aloud 1-2 seconds behind the speaker.
- Shadow the full clip at reduced speed. Run the entire clip at 80-90% of original speed and hold the delay without collapsing into pausing.
- Record and compare. Record your shadow, play it against the original, and compare on exactly three points: stressed words, word linking, and sentence-final intonation.
Accuracy is the gate, speed is the reward
Stay at reduced speed until you can shadow a 60-second clip missing no more than 5 words — a mumbled word counts as missed, so count honestly. Once that count is at or under 5, move to full speed at the same standard. Two habits break the gate: mouthing, because your ear is the only judge and a silent mouth gives it nothing to judge, and jumping to full speed early, because fast mumbling sounds like nothing outside.
Shadow ideas, not syllables
Keep your attention on what the speaker means, not on the shape of individual sounds. Echoing sense transfers to real conversation; echoing syllables falls apart the moment you have something to say. If you catch yourself producing sounds without understanding, stop and replay that line.
A three-week ladder, then a weekly count
Climb by sentence type. Week 1, short statements. Week 2, questions and requests, where intonation does more work. Week 3, longer compound sentences that hold several ideas in one breath. Close each week the same way: record one 60-second shadow and count the words you missed or mumbled. Let the count, not the feeling, decide whether you climb; if the number does not drop week over week, stay on the same rung. For the wider loop — a call, its live subtitles, and per-turn feedback connecting back to what you rehearsed — see the four-step practice loop. For a day-by-day calendar built around calls, the seven-day speaking practice plan lays out the first week.
Where coach audio fits in
YouSpeak runs this exact loop as calls with an AI coach: play a coach line, shadow it under the same rule — 1-2 seconds behind, voice audible — then record and compare. You can practice in 5 conversation languages — English, Chinese, Spanish, French and Japanese — and the app interface itself comes in 18 languages. Every plan in pricing includes daily conversation time, and no card is needed to start; the other guides live in the insights hub.