Background noise test

Everyone can hear that your line is noisy. Nobody can tell you how noisy. This measures the level your microphone hears when you say nothing, then the level when you speak, and reports the gap between them — which is the number that decides whether you sound clear or tiring to listen to.

How the measurement works

Two five-second passes. Stay quiet for the first: that is your floor. Speak normally through the second: that is your voice. The difference between them is the gap.

The floor is reported as the median of the readings rather than the average. One cough during a five-second window moves an average by ten decibels and would report a quiet room as a noisy one; the median ignores it.

All browser audio processing is switched off during the measurement. Otherwise the browser would remove the noise before it could be measured, and the result would describe the browser rather than your room.

What to change, in order of how much it helps

1. Move the microphone closer. Free, and it beats everything else. Halving the distance raises your voice by about six decibels while the room stays where it is. Twenty centimetres from the corner of your mouth is a good target for a headset boom; thirty for a desk microphone spoken across rather than into.

2. Turn the input gain down and get closer, rather than up and stay away. Gain amplifies your voice and the room equally. Distance does not.

3. Take the microphone off the desk. A desk carries every keystroke, every cup, every knee. Anything soft between the stand and the surface removes a class of noise that no software can, because it arrives as vibration rather than sound.

4. Turn off what is running. A laptop fan spins up under load and is often the single loudest thing in a quiet room. Check what is using the processor before blaming the microphone.

5. Change the cable or the port, if it hisses evenly. A steady, tuneless hiss that does not change when the room does is electrical rather than acoustic. USB microphones sharing a hub with a busy device pick this up; so do analogue microphones in a front-panel socket.

Windows 11

  • Turn off input enhancements while measuring. Settings → System → Sound → your input device → Audio enhancements. These often include a noise gate, which will make the floor look far better than it is.
  • Check the level, not just the noise. Settings → System → Sound → Input → Volume. A microphone at 100 with the gain compensating for distance is the most common cause of a bad gap.
  • Exclusive mode — device properties → Advanced — lets one application take the device and change its settings underneath everything else. If numbers move between runs for no reason, check it.

macOS

  • System Settings → Sound → Input. Watch the input level while the room is quiet. If the bar moves noticeably with nobody speaking, the floor is high before any test says so.
  • Ambient noise reduction, where the input offers it, appears in the same panel. Off while measuring.
  • MacBook internal microphones are unusually good at rejecting the room and unusually close to the fans. A result that is quiet in general use but jumps under load is almost always thermal.

When the number is fine and calls still sound bad

Then it is not noise. Two things it is instead, both covered by their own tests here: the level is too low, so the far end amplifies everything including your room; or your speakers are feeding back into the microphone, which sounds like noise but behaves completely differently.

Common questions

What is a noise floor?

The level your microphone reports when nothing is deliberately making a sound. It is never silence: it is the room, the computer fans, the street outside, and the electrical noise of the microphone and its preamplifier added together. Measured in dBFS, so the numbers are negative and lower is better.

What counts as a good number?

Below minus 60 dBFS is quiet — a decent microphone in a normal room with nothing running. Between minus 60 and minus 45 is workable and typical of a laptop in an office. Above minus 45 means the room is arriving at roughly a tenth the loudness of a speaking voice, which the far end will hear as a constant hiss under everything you say.

Why does the gap to my voice matter more than the floor?

Because it is the ratio your listeners actually experience. A noisy room with a close, loud microphone can sound better than a quiet room with a distant one. Forty decibels of gap is comfortable. Below twenty five, noise suppression starts having to damage your voice to remove the noise, which is what produces that underwater sound on calls.

Should I just turn on noise suppression?

It is the last step, not the first. Suppression works by removing what it believes is not speech, and the less gap it has to work with, the more of your voice it takes with it. Lower the noise first and it will barely have to act. This test measures with all browser processing switched off, so the number you see is the real one.

Why does my number change between measurements?

Because rooms are not constant. A fridge compressor, a fan spinning up, traffic — all of it moves the floor by several decibels. Measure two or three times. The test reports the median rather than the average precisely so that one door closing does not decide the result.

Other checks for this device

  1. Microphone test
  2. Echo test