Hear a phrase. Say it back. Until it sticks.

How it works

1. 1

   Load a recording

   A podcast episode, an audiobook chapter, a clip from a show — any audio file your browser can decode. It is read on your device, not uploaded.

2. 2

   Cut it into phrases

   Drop a boundary in each pause the speaker left, or let auto-split place them all. Two to six seconds per phrase is the range that works.

3. 3

   Loop, speak, compare

   Each phrase repeats with a gap. Say it back into the gap, record yourself, and play the two versions against each other.

Everything a session needs

Learn the technique

The tool is the easy part. These cover the method itself — session length, repetitions, choosing material, and what to listen for when you play your voice back.

• What is shadowing? The technique interpreters use to learn languages

  Where the technique came from, what it actually trains, and the honest case for and against it.

• How to shadow: a step-by-step method for your first session

  A repeatable routine: clip length, repetitions, what to listen for, and when to move on.

• How to shadow with podcasts (and which episodes actually work)

  Picking episodes that work, cutting them into phrases, and practising away from a screen.

• Shadowing vs. listen-and-repeat: which one should you be doing?

  Two exercises that look alike and train different things — and when to use each.

Common questions

What is shadowing?

Shadowing means repeating speech aloud while the recording is still playing, trailing the speaker by a beat rather than waiting for a pause. Because the audio never stops, it trains you to produce language under the same time pressure as real conversation — which listening alone does not.

Is Shadowly free?

Yes. Splitting audio into phrases, looping with gaps, adjusting speed, recording yourself, exporting individual phrases as WAV and syncing your splits across devices are all free and need no payment. A subscription adds automatic phrase detection, a library of your files, whole-session MP3 export and ad-free practice.

Are my audio files uploaded to a server?

No. Your recordings are decoded and processed entirely in your browser, and neither the audio you practise with nor the voice recordings you make are ever sent to us. If you create an account, only the metadata — file names and where the phrase boundaries sit — is synced so your work follows you between devices.

Do I need an account to use it?

No. The editor is fully usable without signing in: load a file, split it, loop it and record yourself. An account exists only to sync your phrase boundaries across devices and to carry a subscription.

What audio files can I use?

Anything your browser can decode, which in practice means MP3, WAV, M4A, AAC, OGG and FLAC. Most podcasts publish a plain MP3, which is the easiest place to start.

How long should a shadowing session be?

Fifteen to twenty minutes daily achieves far more than a single long weekend session, because this is motor learning and it consolidates across sleep. Sixty to ninety seconds of audio is enough material for a full session once you are repeating each phrase five to eight times.