Why Your Screen Recording Has an Echo (and How to Stop It)
Echo, doubled voices, and hollow repeats in a screen recording usually come from one loop between speakers and mic. Here's how to find and fix it.
You finish a take, hit play, and your voice comes back twice. Sometimes it is a hollow repeat a fraction of a second behind. Sometimes it is a full second copy, like two people reading the same script slightly out of step. Either way the recording is unusable, and re-recording usually reproduces the same problem, because the cause is in your audio setup rather than in the take.
Almost all of it traces back to one thing: sound leaving your speakers and arriving back at your microphone. Once you know where that loop is, the fix takes about a minute.
Three Problems That All Get Called “Echo”
Before changing settings, identify which one you have. They sound similar but have completely different fixes.
- Doubling. Every word appears twice, offset by anywhere from a tenth of a second to a full second. The second copy often sounds thinner or more distant. This is a loop between your speakers and your microphone.
- Reverb. Your voice sounds like it was recorded in an empty kitchen — one voice, with a tail of room sound after it. Nothing is doubled. This is the room, not the software.
- Feedback. A rising whine or howl during the take. This is a loop that got loud enough to feed itself, usually speakers pointed at an open microphone with the gain turned up.
Doubling is by far the most common complaint, so start there.
Cause 1: Speakers Plus Microphone Plus System Audio
Recorded captures the microphone and system audio as two separate sources. In the main window you have three toggles — Microphone (with a device menu and a live level meter), Camera, and System Audio. When both audio toggles are on, two things are recorded at once:
- Everything your computer plays, tapped directly from the system’s audio output
- Everything your microphone hears
If you listen through speakers, the microphone also hears the speakers. Anything playing — a video in your demo, a conference call, a sound effect — lands in the file twice: once clean from the system tap, and once a few milliseconds later, muffled, through the air into your mic. Your ears ignore this while you record. The playback does not.
The fix is headphones. Any headphones, wired or otherwise. The moment your speakers stop playing into the room, the second copy is gone. This single change resolves most doubled-audio reports.
If you cannot wear headphones, your options are to turn System Audio off when the recording does not actually need it, or to turn the speaker volume down far enough that the bleed sits under your voice. Narration-only tutorials rarely need system audio at all.
Cause 2: You Are in the Call You Are Recording
Recording a meeting is a special case of the same loop, and it catches people even when they are wearing headphones.
When you record a video call with system audio on, you capture what the meeting app plays back to you — which is everyone else. Your own voice should not be in that stream. But if you are also speaking into a microphone that someone else’s setup is re-broadcasting, your voice comes back through the call a few hundred milliseconds late and lands in the system track, on top of your own clean mic track.
You cannot fix this on your end by changing Recorded’s settings, because the delay is introduced before the audio reaches you. Ask the other participant to use headphones, or record your mic track only and accept that their side will be missing. For more on getting usable audio out of a call, see recording remote meetings and interviews.
Cause 3: Your Microphone Is Being Routed Into Your Output
This one looks like a software bug and is actually a setting you turned on months ago.
System audio capture taps the output stream, so your microphone should never appear in it. But some setups deliberately route the mic back into the output so you can hear yourself:
- Windows: the “Listen to this device” option in the microphone’s properties plays your mic through your speakers or headphones. Anything in your output is in the system audio track, so your voice is recorded twice — once from the mic, once from the loopback.
- macOS: audio interfaces with software monitoring, multi-output or aggregate devices, and virtual audio drivers left over from an older recording workflow can do the same thing.
- Both: a DAW, streaming app, or conferencing tool left running with monitoring enabled will push your mic into the output even when its window is closed.
Turn monitoring off, or switch your output back to plain speakers or headphones before you record. Recorded does not need a virtual audio driver on either platform — it uses ScreenCaptureKit on macOS and WASAPI loopback on Windows, both of which read the system output directly. If you installed something like BlackHole for an older tool, it is a likely suspect. The platform walkthroughs cover the clean setup: Mac internal audio and Windows 11 with system sound.
Cause 4: It Is Not Echo, It Is the Room
If there is one voice with a tail rather than two distinct copies, you have reverb. Hard parallel surfaces — bare walls, a glass desk, an empty room — reflect your voice back into the microphone milliseconds later, and the reflections smear together.
Recorded applies no echo cancellation or noise removal; it stores what the microphone sends. That is deliberate — those processes damage good audio — but it does mean room problems have to be solved in the room:
- Get closer to the microphone. Halving the distance roughly doubles your voice relative to the reflections.
- Point the mic away from the largest bare wall and away from the screen.
- Add soft things: a rug, curtains, a bookshelf, even a jacket over a chair behind you.
- Lower the input gain and speak closer rather than turning the gain up and sitting back.
There is more on placement and levels in audio recording best practices.
Catch It Before the Real Take
The whole diagnosis takes thirty seconds:
- Put headphones on.
- In the main window, enable only the audio sources you actually need. Narration-only content needs the microphone. A demo with video playback needs both.
- Watch the meters. The microphone toggle shows a live level while you talk, and the System Audio toggle shows one while something plays.
- Record ten seconds: say a sentence, then play something with sound, then say another sentence.
- While recording, the control window shows separate waveform indicators for mic and system audio. If the system waveform moves while you are talking and nothing is playing, your voice is getting into the output — that is Cause 1 or Cause 3.
- Play the test back before the real take.
What You Can Still Fix in the Editor
Because the microphone and system audio stay separate all the way through, a take with echo is not always lost.
Select a segment in the timeline and open the Segment tab. Each segment has independent System Audio and Mic controls, each with a volume slider and a mute toggle. If the doubling comes from system audio capturing your speakers while you narrate, mute the system track on the segments where only your voice matters, and leave it unmuted over the part where the demo video plays. The timeline waveform helps you find the boundaries — system audio is drawn on the top half, the microphone on the bottom.
What this cannot do is separate two copies that are already mixed into one track. If your voice reached the microphone twice, both copies are in the mic recording and no amount of volume adjustment will remove one of them. That take needs to be re-recorded, or cut out — see trimming and cutting screen recordings.
The Short Version
Wear headphones. Turn System Audio off when the recording does not need it. Check that nothing is routing your microphone into your output. Record ten seconds and listen before you commit to a twenty-minute take.
If you are setting this up for the first time, download Recorded, run the ten-second test, and you will know within a minute whether your audio path is clean.