DevicedialWhy is my microphone so quiet?
A microphone peaking below −20 dBFS in the Devicedial mic test is quiet; the level to aim for is peaks between −12 and −6 dBFS while you talk normally. The scale is negative because 0 dBFS is the ceiling where the signal clips, not silence: −20 dBFS is a tenth of full-scale amplitude, −12 is a quarter and −6 is a half.
Four causes account for nearly every quiet microphone: a low input level in the operating system, sitting too far away, automatic gain control pulling the level down, and another program holding the device in exclusive mode. Distance is the one people underestimate — in the open, sound pressure falls about 6 dB every time the gap between mouth and capsule doubles, so 40 cm instead of 20 cm costs you half your level. A microphone that reads fine here but goes silent in a call is not broken: that is a device-selection or permission problem, and it has a different fix.
How quiet is actually too quiet?
Below −20 dBFS. That is the line the mic test draws, and it is drawn there because of what happens next rather than because the number sounds bad. A signal peaking at −20 dBFS is a tenth of full-scale amplitude, so anything downstream — a call app, a recorder, the other person’s headphones — has to add roughly 14 dB to bring you up to a comfortable level, and it adds that to your keyboard, your chair and your neighbour’s lawnmower at the same time.
Peaks between −12 and −6 dBFS are the target. At −6 you are at half of full scale with 6 dB of headroom left, which is enough for a laugh without clipping. The reason to talk in peaks rather than in an average is that clipping is a peak event: an average that looks polite can still be slamming into 0 dBFS several times a minute, and once a sample is clipped there is nothing to recover.
Why is it quiet when the level slider is already high?
Because the slider is only one of four things in the chain, and it is usually not the one that is wrong.
Distance. Sound pressure in the open falls about 6 dB every time the distance doubles. Moving from 20 cm to 40 cm halves your level, and a laptop microphone at arm’s length starts a long way behind a headset boom that sits 3 cm from your mouth. This is the cheapest fix available and it is almost always the largest one.
Automatic gain control. Browsers and call apps apply it by default. It lifts you when you go quiet and pushes you down when you get loud, which flatters speech and ruins measurement — if the number moves while you hold a steady note, you are watching the software, not the microphone.
The wrong device. A machine with a webcam, a headset and a built-in array has three inputs, and the one the system calls “default” is rarely the one you meant. Pick devices by name everywhere.
Something else holding the device. On Windows an application can take an input in exclusive mode, and while it does, the rest of the system gets silence rather than a quiet signal. That is the next section.
Why does the mic work here but not in Discord or Teams?
Because the browser and the desktop app are asking different questions of the operating system. The mic test requests the default input and shows you what arrives. A desktop app keeps its own device choice, its own gain, and its own system-level permission — three settings that can each be wrong while the hardware is perfect.
The order that resolves it fastest: confirm the microphone reads a healthy level here, then open the app’s audio settings and select the device by name, then check the operating system’s microphone privacy list to make sure that particular application is allowed to record at all. On Windows that list sits in Settings under Privacy, on macOS under Privacy & Security. A camera that behaves the same way has the identical cause, which is why the camera test is worth running alongside when a whole headset seems to have vanished.
What is exclusive mode, and how do I get my microphone back?
Windows audio runs shared by default: the mixer takes every application’s stream and combines them. Exclusive mode is the alternative, in which a single application takes the endpoint and bypasses the mixer entirely — useful for low-latency recording and bit-perfect playback, fatal for everything else that wanted the same device.
The symptom is unmistakable once you know it: exactly one program can hear the microphone, and every other program hears nothing at all, including this page. The fix is in the device’s Advanced properties, where two checkboxes allow applications to take exclusive control and give them priority. Clear both, then restart the application that was holding the device. The same pair of checkboxes on the output side is worth clearing if your speaker test or headphone test has gone silent for one program only.
What order should I fix things in?
Closest thing first, cheapest thing first — they happen to be the same list.
- Move the microphone nearer, or move yourself. About 6 dB per halving of distance, free.
- Check the input level in the operating system, not in the app.
- Turn off automatic gain control if you are measuring rather than talking.
- Select the device by name in every application that matters.
- Clear exclusive mode if one program has taken the device hostage.
- Only then reach for microphone boost, remembering that it amplifies the room too.
Re-run the mic test after each step rather than at the end. It takes seconds, and it is the only way to find out which of the six actually did something — the whole reason the test reports a number in dBFS instead of a green tick.
Frequently asked questions
My microphone works in the test but not in Discord or Teams — why?
Because the two are not listening to the same device. A browser test uses whatever the operating system calls the default input, while a desktop app keeps its own input setting, and the two drift apart the moment you plug in a headset. Open the app’s audio settings and pick the device by name rather than leaving it on Default. If the list is empty, the app has not been granted microphone permission at the system level, which is a separate switch from the browser’s.
What is exclusive mode and why does it take my microphone?
Windows audio can be opened in shared mode, where the mixer combines every application, or in exclusive mode, where one application takes the endpoint device outright and the mixer is bypassed. Exclusive mode exists for low-latency and bit-perfect audio work, and while it is held, everything else gets silence. If one program can hear the mic and no other can, clear the two exclusive-mode checkboxes in the device’s Advanced properties and reconnect the apps.
What dBFS should my microphone peak at?
Peaks between −12 and −6 dBFS when speaking at your normal volume. That leaves 6 to 12 dB of headroom for a laugh or a raised voice, which is what stops the loud moments clipping at 0 dBFS. Below −20 dBFS the signal is quiet enough that whatever the other end does to make you audible will amplify your room as well as your voice. Peaks touching 0 dBFS are worse than quiet, because clipped audio cannot be repaired later.
Does the browser change my microphone level on its own?
Yes, unless it is told not to. Media capture applies automatic gain control, noise suppression and echo cancellation by default, and automatic gain control in particular will raise a quiet signal and duck a loud one while you speak. It is the right default for a call and the wrong one for a measurement, so a level that drifts while you hold a steady tone is usually the browser, not the microphone.
Should I turn on microphone boost?
Only after the physical fixes. Boost is gain applied after the capsule, so it lifts the room noise, the fan and the desk bumps by exactly as much as it lifts your voice — the signal-to-noise ratio does not improve. Move closer first, since halving the distance is worth about 6 dB for free, then raise the input level, and use boost to close whatever gap is left.
How the test works
Sources: