Noise suppression
What is noise suppression?
Noise suppression is audio processing that removes steady background sound such as fans, keyboards and traffic from a microphone signal while keeping the speaker's voice. Modern versions use a small neural network that runs on your own device, so what leaves your computer is the cleaned-up voice rather than the whole room.
How does noise suppression work?
The microphone signal is processed in tiny slices, a few milliseconds at a time, before it is sent. Classic suppressors estimate the steady background hum and subtract it. AI suppressors use a model trained on large collections of recorded speech and noise; for each slice, it estimates which parts of the sound are voice and turns the rest down. Because it reacts slice by slice, it can also catch short sounds such as typing or a dog barking.
Noise suppression is one of three audio clean-ups in a call, and each one solves a different problem.
| Processing | What it removes or fixes |
|---|---|
| Noise suppression | Background sound around you: fans, traffic, keyboards, a vacuum next door. |
| Echo cancellation | Other people's voices coming out of your speakers and back into your microphone. |
| Automatic gain control | Volume swings, so you sound equally loud near or far from the microphone. |
Why does noise suppression matter in a call?
Constant background noise is tiring to listen to and makes speech harder to follow, especially for someone listening in a second language. It also matters for anything that turns speech into text: a cleaner signal gives a transcript fewer errors.
It has limits. Aggressive suppression can clip quiet words or the ends of sentences, it treats music as noise, and the processing uses some of the computer's power, which is why calling software may turn it off on weaker devices. It does not fix a bad microphone placement; moving the microphone closer usually helps more.
How does noise suppression work in Meeto?
In the browser, Meeto has an AI noise suppression switch labelled “Noise suppression (AI)”. It sits in the call's settings, in the Devices section: on a computer open More, then “Settings & devices”; on a phone tap the “⋯” button, then “Settings”. It is on by default on desktop computers that have the processing power for it, and off by default in phone browsers, where you can turn it on. Your choice is remembered in that browser for later calls.
The processing runs on your device. When the switch is off, the browser's own lighter processing stays on, and echo cancellation and automatic volume control are always on. If the AI processing fails, the call falls back to the plain microphone without interrupting you. In the iPhone and Android apps, the phone's built-in voice processing is always on, and there is no separate switch.
If people hear their own voice repeated, that is echo rather than noise: see how to fix echo. If your microphone is not picked up at all, start with the device troubleshooting guide, and see virtual backgrounds for the video counterpart.
Frequently asked questions
Should I turn noise suppression off to play music?
Usually yes. Suppression is tuned for speech and can treat instruments or singing as noise, cutting notes or pumping the volume. In Meeto, switch "Noise suppression (AI)" off; the browser's own lighter processing stays on.
Does noise suppression fix echo?
No. Echo is your own voice coming back through someone else's speakers and microphone, and it is handled by echo cancellation on their device. The echo guide explains how to find the participant causing it.
Will it remove other people talking in my room?
Mostly not. Other voices are speech too, so a suppressor keeps much of them. A headset microphone close to your mouth, or a quieter room, helps more than any filter.
Is my audio sent somewhere to be cleaned?
Not in Meeto. In the browser the noise suppression runs on your own device before the audio is sent; in the apps the phone's built-in voice processing does the same.
Try your first browser meeting
Create a room, share the link and invite someone to join you.