SoundForgePro Official 15-day trial

Audio Tools

Online Microphone Test: Level, Clipping and Noise Check

Guide navigation Table of contents Sections
    Studio microphone feeding a waveform, level meters and spectrum display for an online input check

    Run this test before you open a recorder or start changing drivers. It answers one narrow question: is the browser getting a clean enough signal from the microphone you plan to use? Press Start, speak at the distance and level you'll actually record at, then make a short sample and listen back.

    Permission is requested only after you press Start. Audio stays in this tab.

    How to read the microphone meter

    The numbers are digital levels in dBFS. Zero dBFS is the digital ceiling; it has nothing to do with acoustic sound pressure. A peak near 0 dBFS means the converter or browser stream has almost no headroom left. It tells you nothing about how loud the room is in dB SPL.

    ResultWhat it meansNext move
    Peak above −1 dBFSThe stream is at or very near the digital ceilingReduce interface or system input gain and repeat the loudest line
    Peak from roughly −12 to −6 dBFSA practical spoken-word starting range with useful headroomListen to the sample for room tone, handling noise and processing
    RMS stays below about −45 dBFS while speakingThe selected input may be too quiet, too far away or mutedConfirm the device, distance and gain before boosting later
    Meter moves in silenceYou are seeing the current input chain, including gain and browser processingMute fans and HVAC, move the mic, or lower unnecessary gain

    Treat these ranges as working guides rather than a delivery standard. A drum hit, whispered narration and a video call need different margins. What you're avoiding is clipping on one side and a recording so quiet that later gain drags the noise up with it on the other.

    Microphone signal shown at low, healthy and clipped recording levels
    A useful input check separates a signal that is too quiet from one with working headroom and one already flattened by clipping.

    Run a useful 30-second check

    1. Select the real input after permission is granted. Device labels may be hidden before the browser has permission.
    2. Say a quiet phrase, a normal phrase and the loudest line you expect to record.
    3. Watch the peak as well as the moving average. A short overload is easy to miss by ear.
    4. Record a short sample and listen on headphones. Check for plosives, hum, room echo, keyboard vibration and automatic gain changes.
    5. Stop the test before changing cables, interfaces or microphones. The Stop button ends every media track and the live analysis.

    Why the noise reading is only an input-chain estimate

    The page measures the samples the browser delivers. It can't calibrate the microphone, preamp, interface or room. Input gain changes the reading, and so do automatic gain control, echo cancellation and noise suppression. The settings line shows those values whenever the browser exposes them.

    Use the quiet reading to compare one controlled change at a time: fan on versus off, microphone close versus far, laptop on the desk versus isolated. Just don't call the result the room's acoustic noise level.

    If the meter is silent

    Start with the selected device. A laptop may offer a built-in array, a webcam microphone and an interface all at once. If the right name is selected and the meter still doesn't move, check the hardware mute, the interface gain and the operating system's input level. Close any communication software that might be holding the device in exclusive mode.

    If the browser says permission is blocked, change the microphone permission for this site in the browser's site settings. Pressing Start over and over won't override a stored denial.

    If the signal clips

    Lower gain at the earliest controllable point. If the microphone or interface clips before the signal reaches the browser, pulling a fader down later only makes the clipped waveform quieter. Repeat the loudest phrase after every gain change and leave margin for a performance that is stronger than the rehearsal.

    When the problem is a burst on P or B rather than broad overload, use the workflow in the plosive-removal guide; prevention with placement and a pop filter is usually cleaner than repair.

    Privacy and browser support

    The test calls the browser's getUserMedia() method only after Start. The live waveform comes from an AnalyserNode. The optional sample uses MediaRecorder and stays as a temporary Blob in this tab. There is no upload step.

    Microphone access requires HTTPS and a browser that supports the relevant media APIs. Stop, Discard, reload and closing the tab all remove the in-memory sample; Stop also ends the microphone tracks.

    What this test cannot diagnose

    It can't identify a damaged capsule, measure frequency response, certify a noise floor, calculate dB SPL or prove that a Sound Forge recording route is set up correctly. For editor-specific input and driver problems, use the Sound Forge recording troubleshooting guide. This page is the quick browser-side check you run before that deeper path.