Slapback · 2026-09-20
Echo vs reverb: why the difference decides your fix
"There's echo on the call" is doing double duty in English, and it costs people real troubleshooting hours. Under that one word live two different physical events with two different fixes — and if you chase the wrong one, every setting you touch will be the wrong setting. Here's the split, and the ten-second test that tells you which side you're on.
Echo: a copy, arriving late
Echo — the thing this site is named after — is a discrete repetition. A voice happens once, and then it happens again: distinct, delayed, recognizable. On calls, the copy is manufactured by a round trip — a voice leaves a speaker, enters a microphone, and gets re-transmitted, arriving hundreds of milliseconds after the original because networks take time. Your ear hears two events. That gap is the fingerprint: if you can point at the moment the repeat starts, you have echo.
Audio engineers even name echoes by their gap. A single tight repeat — the kind rockabilly producers put on vocals — is called slapback, which is where this blog got its name. On a call, though, nobody's doing it for style: a slapback of your own voice means a device somewhere is looping, and the fix is finding it (whose loop is it?) and breaking it (without muting everyone).
Reverb: a smear, not a copy
Reverb is what a room does to every sound made inside it: thousands of reflections off walls, ceiling, glass and desk, arriving so fast and so densely that the ear can't separate any of them. You don't hear a repeat — you hear a tail, a hollowness, the "empty apartment" ring. Nothing made a round trip; the room itself smeared the sound before the mic ever transmitted it. On calls, reverb means your voice arrives sounding distant and boomy, but nobody ever hears anything twice.
The distinction has a clean physical line: reflections merging within about a twentieth of a second read as one smeared sound (reverb); a reflection or re-transmission arriving later than that reads as a separate event (echo). Calls put network delay on top, which is why call echo is so unmistakably two sounds.
The ten-second test
Ask one question: can you count the voices? "I hear the sentence, then I hear it again" → echo. Someone's device is looping; no room is guilty; go hunt the device. "The voice sounds like it's at the bottom of a well, but only once" → reverb. The speaker's room is guilty; no device is looping; nothing you do to your setup will help their room.
Why the fixes never cross over
- Echo fixes: break the speaker-to-mic loop on the looping device — volume down, mic moved, duplicate device evicted, acoustic echo cancellation reinforced with a dedicated layer if the setup demands open speakers. Soft furnishings won't fix an echo: the loop is electrical-acoustic, not decorative.
- Reverb fixes: get the mic closer to the mouth (the single biggest lever — close voice drowns the room), speak toward the mic, soften the room within reason, or run software de-reverberation — the job Krisp's Room Echo Cancellation and NVIDIA Broadcast's Room Echo Removal are named for. Muting other participants won't fix reverb: there's no loop to break, the room rings all by itself.
Vendors bundle both jobs under "echo" because that's the word people search — which is fair, but it means reading feature lists carefully: acoustic echo cancellation and room echo removal are different machinery, and a tool can carry one without the other. Our software guide labels which tools do which half, and the Krisp post walks the two features by name.
One last boundary, for completeness: a voice repeating instantly in your own headset is neither of these — that's sidetone, a monitoring feature, and it has its own page. Three sounds, three mechanisms, three fixes. The ear that can tell them apart never wastes an afternoon in the wrong settings panel again.