Measure sound level from the microphone. The meter applies a weighting curve and a time constant the way a sound level meter does, then reports the level, a running Leq, and the statistics that describe how much the sound moves.
Calibration is the whole question
A microphone produces a digital level. Nothing in that number is tied to pressure in the room, and the relationship differs between a laptop, a headset, and an interface with a measurement capsule.
The offset closes that gap. It is added to the digital reading to produce a sound pressure level, so it must be set once per device against something you trust. A calibrator producing 94 dB at 1 kHz is the accurate route. Matching a handheld meter or a phone in the same position is rougher, and still far better than the default.
Without calibration the numbers are relative. That is not useless: a room that reads 12 dB higher with the ventilation on is 12 dB louder either way, and the comparison holds even when the absolute figure does not.
A, C, and Z weighting
Hearing is not equally sensitive across frequency, and at ordinary levels it discounts the bass heavily. A weighting is a filter that mimics that, which is why almost every noise regulation is written in dB(A).
C weighting is nearly flat from 30 Hz upward. It is used for peak measurements and for loud, low sound, where the ear is closer to flat and A weighting would ignore energy you can feel.
Z is no weighting at all, the raw level across the band. Use it when you want the signal rather than an impression of it.
The difference between the A and C readings on the same sound is a quick description of its spectrum: a large gap means most of the energy is low.
Time weighting
The reading is an average, and the time constant sets the window.
Fast averages over 125 ms, which tracks speech and traffic. Slow averages over 1 second, which is steadier and easier to read in a room whose level keeps moving. Impulse rises in 35 ms and decays slowly, so a door slam or a hammer blow stays on screen long enough to read rather than flicking past.
None of these change the underlying sound. They change how much of it the number represents at any moment.
Leq, Lmax, L10 and L90
A single reading describes an instant. The statistics describe a period.
Leq is the equivalent continuous level: the steady sound that would carry the same energy as everything since the last reset. It is the figure limits are written against, because 5 minutes of machinery and one loud minute can add up to the same exposure.
L10 and L90 are the levels exceeded 10% and 90% of the time. L90 is a good stand-in for the background, since it is what remains once the events are stripped out, and L10 describes the events themselves. A wide gap means a room dominated by occasional noise; a narrow one means steady noise.
Typical levels for orientation
A quiet bedroom at night sits around 30 dB(A). A library or a quiet office is near 40. Conversation at a metre is 60 to 65. A busy street is 80, and a power tool at arm’s length is over 100. Sustained exposure above 85 dB(A) is where hearing protection starts being recommended.
These are orientation figures, not a standard, and they are worth knowing mainly as a sanity check on a calibration.
Where a browser meter stops
A phone or laptop microphone is built for voice. It rolls off in the bass, often has its own processing before the browser sees anything, and compresses or distorts well before a real measurement microphone would. The meter disables automatic gain control, echo cancellation, and noise suppression, since all three move the quantity being measured, but it cannot undo what the hardware does.
The practical limits: very quiet rooms hit the microphone’s own noise floor, and loud events run into the input clipping long before the number on screen suggests. For anything that has to stand up, a class 1 or class 2 meter is the instrument, and this is the one that tells you whether that trip is worth making.