Measure peak/RMS in dBFS. For gain-staging before streaming or recording.
Your audio is never sent outside this device. Measurement happens entirely inside your browser.
When you press the button, your browser will ask "Allow this site to use your microphone?"
This page disables echo cancellation, noise suppression, and auto gain entirely, so it measures the raw, unprocessed signal.
Microphone access has not been granted.
Click the 🔒 (or microphone icon) in the address bar, set this site's microphone permission to "Allow," and reload the page.
No microphone was found.
Please check that a microphone is connected.
Your browser doesn't support measurement.
Please try the latest version of Chrome, Edge, Safari, or Firefox.
Couldn't start the microphone.
If another app is using the microphone, close it, then reload the page and try again.
-60-30-18-12-60 dBFS
Healthy (aim for -18 to -12)
Caution (-12 to -6)
Too hot (above -6 risks clipping)
Peak (Sample Peak): the amplitude of the loudest sample within the measured interval. 0 dBFS is the digital ceiling — reaching it causes clipping.
RMS: the average loudness of the audio. It's closer to how loud a voice "sounds" to the ear. If there's a large gap between peak and RMS, the mic may be too far away or the room may be too reverberant.
For streaming/recording gain-staging, a common guideline is to keep the peak from exceeding roughly -12 to -6 dBFS at the loudest voice you might actually produce (exact standards vary by format and situation).
This page measures the signal with echo cancellation, noise suppression, and auto gain all disabled. Numbers may differ from a typical mic test page.
What's measured is the amplitude of the digital signal that reaches your browser. It cannot measure mic preamp distortion, OS-level boosting, or actual sound pressure (dB SPL).
Mic adjustment steps (common prep)
Temporarily disable all effects in your streaming software or OS — noise suppression, compressor, auto gain, etc. (otherwise you can't see the raw input level).
Fix the distance between the mic and your mouth at the position you'll actually use (level changes with distance, so decide the position first).
Start measuring on this page and speak both at your normal voice and at the loudest voice you might actually produce, checking the peak (max).
Adjust your OS mic input level (or your audio interface's gain knob) so it fits the target guidelines for your use case below.
Once adjustment is done, restore the effects you disabled.
Guidelines by use case
🎙 Talk streams / web meetings
Adjust so your normal voice peaks around -18 to -12 dBFS.
Ideally, even momentary spikes like laughter shouldn't exceed -6 dBFS. If they do, lower the input level a bit.
If RMS is dramatically lower than peak (a gap of 20 dB or more), the mic may be too far away or picking up too much room reflection. Try moving it closer to your mouth.
🎮 Streams with loud voices or shouting (e.g. game commentary)
Base your setting not on your normal voice but on your loudest shout. Actually shout, and set the input level so that peak doesn't exceed roughly -6 dBFS.
As a result, your normal voice will sit in a modest -24 to -18 dBFS range — that's expected and normal. It's standard practice to boost the quieter parts afterward with a compressor or gain in your streaming software.
If you set the level based on your normal voice first, a sudden shout will hit 0 dBFS and clip. The "max" display is useful for capturing your loudest shout.
🎤 Singing (song covers, singing with an instrument)
Actually sing the strongest phrase in the song (e.g., the highest note in the chorus) and adjust so its peak lands around -12 to -6 dBFS. Keep it more conservative than talk to leave headroom for post-recording processing (mixing).
Check the "max" for both quiet and strong phrases, and prioritize making sure the strong phrase doesn't exceed the limit.
If your mic technique involves changing distance during the song, measure at the closest position you'll use.
If pop noise (plosive bursts from "p" sounds) causes momentary peak spikes, address it with a pop filter or mic angle rather than by lowering the level.
The input level cannot be changed from this page. Adjust it via your device's microphone settings (or your audio interface's gain knob).
* These are general guidelines. The ideal target varies by streaming platform, recording environment, and downstream processing.
* If you want to listen to your own voice by ear, use "Hear yourself" on micsound Mic Test (this page is measurement-only and never plays audio back).
Frequently asked questions
Q. The reading barely moves, or stays near −∞ or 0.
A. First check under "Measuring microphone" that the selected mic is actually the device you want to use. If it still doesn't move, check whether your OS input volume is at 0, or whether another app is currently holding that microphone. On some browsers or with certain extensions, noise suppression or auto gain control can be forced on even when this page requests it be disabled, which can make the reading look unnaturally steady compared to your actual volume.
Q. Why Sample Peak instead of True Peak?
A. True Peak (which estimates the actual peak that can occur between samples via interpolation) requires dedicated oversampling processing that the browser's standard APIs can't provide accurately. This page shows the raw sample values obtainable directly via the Web Audio API (Sample Peak). Since Sample Peak can read slightly lower than True Peak, leave a few dB of headroom when fine-tuning levels near the limit.
Q. How is this different from the "loudness normalization (LUFS)" used by YouTube or Spotify?
A. They're different metrics. The dBFS this page measures is the instantaneous level of the input signal, used for gaining your mic (avoiding clipping or too-quiet audio). LUFS (loudness), on the other hand, is a time-averaged measure of perceived loudness during playback, used by streaming platforms for automatic volume normalization. Recording or streaming with proper input gain does make downstream loudness processing work better, but you can't apply this page's reading directly as a LUFS target.