Shadowing English: how to do it, and the best apps and audio to do it with

Shadowing trains your mouth, not your knowledge. You play a short piece of natural speech and repeat it a beat behind the speaker, without stopping to think, copying stress, rhythm and the places they run words together. After a few weeks the shapes stop being foreign and start coming out on their own.

That is worth doing because the gap most learners have is not vocabulary. You can read a sentence and understand it perfectly, and still not be able to produce it at speed, because producing it is a physical skill your mouth has never practised. Shadowing is the drill that closes exactly that gap and almost nothing else.

What it does not train is deciding what to say. You are repeating someone else's sentences, so you never have to choose a tense, hunt for a word or recover from a wrong one. That is the other half of speaking, and shadowing will not touch it. The people who plateau are usually the ones doing only one of the two.

Pick thirty seconds, not five minutes. Something where you understand nearly everything already. If you are decoding meaning, you have no attention left for rhythm, and rhythm is the whole point.

Listen once without speaking. Just hear where the stresses fall and where words run together. Most of what you will copy is in the gaps, not the words.

Shadow it with the transcript in front of you. Start a beat behind and do not stop when you fall off. Falling off and rejoining is the skill.

Then shadow it without the transcript. This is where it gets hard and where it starts working.

Record yourself once and play it next to the original. Thirty seconds of this teaches more than ten minutes of repeating blind, and it is the step nearly everyone skips. Your ear is the tool; the app is optional.

Ten minutes a day on one clip beats an hour on six. Stay with the same thirty seconds until it feels boring — boring is the sound of it being learnt.

Choosing audio that is too hard. If you can only catch 60% of it, you are doing listening practice, not shadowing. Go easier than feels respectable.

Reading aloud instead of shadowing. Reading a transcript at your own pace is a different exercise with a different benefit. Shadowing means the speaker sets the pace and you keep up.

Never recording yourself. Without the comparison you are rehearsing your own accent at speed. The recording is not optional decoration; it is the feedback loop.

Doing only shadowing. It is a drill, not a method. It makes your mouth fast and leaves your decisions slow, which is why it pairs naturally with actual conversation.

The material matters more than the app. What you want is natural speed, a transcript, and something short enough to repeat twenty times without hating it.

Podcasts with transcripts are the classic answer and they are free. YouGlish is the fastest way to hear one specific phrase said by dozens of real speakers, which is unusually good for the sentence that keeps tripping you. EnglishCentral builds the same loop around video with line-by-line repetition, if you prefer watching to listening.

Graded audio has one advantage over podcasts: it is pitched at a level and it comes with the transcript already aligned. The B2, C1 and C2 listening tests on this site are four-minute conversations at exam pace with the full transcript underneath, free and with no account — originally built for exam practice, and they work as shadowing material without modification.

You want to know which sound is wrong. ELSA, and it is not close. It scores individual phonemes and tells you where your mouth is going, and its Shadow Coach mode is built for exactly this drill. What it will not tell you is whether the sentence you said was the right sentence — it is listening to the sound, not the grammar.

You want a human to hear you. Speechling, though note the catch: the free tier's ten monthly coaching sessions exclude English. For English you are on the paid plan from the start.

You want the exact phrase you are stuck on. YouGlish, free, no account, and you can hear the same phrase in six accents in a minute.

You want the other half. Shadowing makes the shapes automatic; it does not make you choose the right words under pressure. That needs a conversation where somebody is waiting for your answer and tells you when the sentence came out wrong — not the sound, the sentence. An AI you can talk to does that at a volume no human will, and it is the natural partner to ten minutes of shadowing rather than a replacement for it.

Every figure in the table was read on each tool’s own pricing page on the date above. Prices and free tiers move; this table is re-checked and re-dated when they do.