29 August 2026
If you hear an echo, it is almost never your fault — it's the other person's speaker feeding their own audio back into their microphone, which sends it to you a beat late. The fix is nearly always on their end, and headphones end it in one move. Everything else is diagnosis: figuring out whose setup is the loop before you waste ten minutes muting people at random.
Say you're talking. Your voice comes out of the other person's speaker. Their microphone picks it up — because it's an open room, not a closed circuit — and sends it back into the call. You hear your own voice, delayed by however long the round trip took: encode, network, decode, playback. That delay is what makes it sound like an echo instead of just louder audio. Close the loop and it repeats, softer each time, which is why bad echo often sounds like a fading copy of you talking over yourself.
This is why the person who hears the echo is the one being echoed, not the one causing it. The person whose mic is picking up their own speaker can't hear the problem — from their seat, their speaker and their voice and the call all sound completely normal. They have no signal that anything is wrong. You have to tell them.
Skip guessing and ask two questions on the call.
Once you have an answer, mute them and speak — if the echo stops, you've confirmed it, and the fix is theirs to make, not yours.
Headphones remove the acoustic path entirely. Sound that plays into your ears never reaches your microphone, so there's nothing for the mic to pick up and re-send. It's not a setting or an algorithm — it's a physical fact about where the sound goes. This is why 'just wear headphones' fixes more meetings than any software update ever will, and why it's the first thing worth asking rather than the last.
Every video platform runs acoustic echo cancellation (AEC) by default. It works by comparing the audio the device just played out through its own speaker against what the microphone picks up a moment later, and subtracting the match. Done well, it removes a clean, single, predictable echo almost completely — one speaker, one mic, one room, consistent volume.
It gives up in a few specific situations: when the speaker volume is loud enough to distort before it hits the mic, when there's a second microphone in the room picking up a different delay, when the room has enough hard-surface reverb that the 'echo' is really dozens of tiny reflections instead of one clean copy, or when someone's on a laptop with an external speaker plugged into a different audio path than the built-in mic expects. AEC is tuned for the common case — it's not modeling your actual room.
The case that defeats every configuration: two people in the same physical room joined to the same call on two separate devices. Each device's speaker feeds the other device's microphone, and because they're two independent audio systems, no single AEC engine has visibility into both sides of the loop to cancel it. Turning up cancellation, changing input devices, switching platforms — none of it touches this, because the problem isn't in any one device's audio pipeline. It's between them.
The only real fixes: one person mutes their mic and lets the other carry both voices (works if you're close enough to a shared mic to be heard), or one of you leaves the room's call entirely and dials in from elsewhere, or you share a single device and mic for the room instead of joining twice. This case comes up constantly in offices with pod-based hot-desking, and it's worth having as a standing rule rather than rediscovering it every time.
An echoing mic doesn't just annoy people live — it also degrades whatever gets captured. AVAY, like other platforms, transcribes the call as it happens, and a looping echo means the same sentence sometimes shows up doubled or garbled in the transcript, because the audio genuinely contains two copies of it at different delays. Fixing the echo at the source — muting the loud one, moving to headphones — cleans the record as well as the room. It's not a transcription bug; it's the audio itself being sent twice.
| Cause | Who notices it | Fix | |
|---|---|---|---|
| Single person on speakers, no headphones | Everyone else on the call | They switch to headphones or earbuds | — |
| Speaker volume too loud, distorting into mic | Everyone else on the call | They lower speaker volume or move mic further from speaker | — |
| Reverberant room (hard floors, bare walls) | Everyone else, especially on loud passages | Soft furnishings help; headphones bypass it entirely | — |
| Two people, same room, two devices | Everyone else, and it's unfixable by settings | One mutes, or joins from elsewhere, or shares one device | — |
Because it's not your device causing it — your microphone is picking up your own voice cleanly, with no delayed copy to hear. The echo exists on the other end, where your voice is being played through their speaker and re-recorded by their microphone. That's why the fix has to come from them, even though you're the one who's annoyed.
It fixes it for that speaker only while they're muted, which is a workaround, not a solution. As soon as they unmute and talk, the loop starts again if the underlying cause — speaker feeding mic — hasn't changed. Headphones or a proper fix on their end is what actually ends it.
Louder speaker output means more sound energy reaching the microphone, which increases both the volume of the echoed copy and the odds it clips or distorts before echo cancellation can model it cleanly. A quiet room with a loud speaker often echoes worse than a loud room with a quiet one.
No — a better microphone can pick up voice more clearly, but it will pick up the speaker's audio just as clearly too. Echo is about acoustic path, not mic quality. A cheap headset beats an expensive standalone mic every time for this specific problem.
They're related but not identical. Echo is a delayed, recognizable repeat of speech; feedback is the runaway high-pitched squeal that happens when a loop keeps amplifying itself past the point of just repeating words. Echo left unresolved can sometimes escalate into feedback, but most call echo never gets that far.
Echo is almost always the other person's speaker feeding their own microphone, not yours — ask whose voice is looping and whether they're on speakers, and headphones ends it in one move that no setting can replicate.
AVAY is a video meeting platform that transcribes the call itself — no bot joins, because there is nothing to join. Start one at avay.ai, read how each part works in the documentation, or see what it costs.