Echo (shadowing)

Shadowing means saying a line right after the original. Echo breaks that into a small loop you can repeat.

One round of Echo

1. Turn on Echo (E). The current line becomes the Echo region; playback pauses at the line end and rewinds to its start instead of moving on.
2. Listen to the original; slow it down if needed (Shift+,).
3. Press R to record, say the line, press R again to stop.
4. G replays your take, P shows the pitch contour, V shows the pronunciation score.
5. Not happy? Record again — every take is saved.

Try it · shadow one line1. Listen2. Record3. Compare

I'd rather stay in tonight./aɪd ˈræðər steɪ ɪn təˈnaɪt/

Hear originalRecord

Press “Hear original” to start.

Adjusting the region

If a line is too short or the boundary is off, fine-tune it:

Action	Key
Extend the start backward	[
Extend the end forward	]
Shrink the start	{
Shrink the end	}

Clicking another line re-centres the region on that line.

Several lines at once: loop the whole sentence

Subtitles often split one complete sentence into two or three lines (a line break, an interjection). The echo region can span multiple lines: click the dashed “Add to loop” row at the top or bottom edge of the loop (or press [ / ]) to pull a neighbouring line in.

The added lines then loop as one piece — the whole sentence plays in sequence, and recording and scoring cover the full sentence, so you practise how the sentence is said rather than a few orphaned half-lines. Added too much? Shrink back with { / }.

Tip

Intonation crosses line breaks — a rising first half and falling second half often get split by the subtitle. Loop the whole sentence and the pitch contour will reflect your real intonation.

The pitch contour

Press P to open the pitch contour: blue is the original, violet is you — drawn on top of each other, the intonation gap is immediately visible.

How it works: analysis runs entirely on your device. The echo region’s audio is decoded to PCM, pitch is extracted frame by frame with Essentia’s PitchYin family of algorithms, producing “pitch over time” curves; your recording goes through the same pipeline, and both curves are aligned in time and overlaid. It is an acoustic measurement, not an AI score — there is no pass/fail tier, only “how close”.

How to read it, and what it’s worth:

• Closer lines = closer intonation. The overall score tells you how good; the contour tells you where it went wrong — which word should have risen but didn’t, where you dropped too early;
• Practise intonation, not words: per-word pronunciation belongs to the word scores; the rise, stress and rhythm of the whole sentence belong to the contour;
• Trends over single takes: record the same line a few times and watch the violet line converge on the blue — far more useful than obsessing over an 82 vs an 84.

The extension (YouTube / Netflix) difference

Your recording line is always available (microphone recording, analysed locally). The original line needs the video’s audio: YouTube works; Netflix and other DRM-protected platforms can’t hand out their audio, so the original line is unavailable and the chart shows only your recording. For both lines, open the material in the web app or Enjoy Player.

Tips

• Get accuracy slowly first, then speed up for fluency.
• Use the pitch contour for intonation: blue is the original, violet is you — the closer, the better.
• Recording needs microphone permission; the first time you’ll be asked. See the FAQ.