Ohmforge Sound Level (WAV → Leq, LAeq, Lmax)

What it is for

You recorded a noise — a heat pump, a party, a workshop machine — and want numbers. This tool reads a WAV file and computes what a sound level meter would show: the energy average (Leq) with A, C and no (Z) frequency weighting, the maximum levels with Fast (125 ms) and Slow (1 s) time weighting, the minimum, the C-weighted peak and the statistical levels L10, L50 and L90.

It uses the same engine as the sound level meter in Ohmforge's Pocket instruments. Its WAV download gives the same numbers here.

A phone or laptop microphone is not a calibrated instrument. Without calibration the levels are relative to a full-scale sine (0 dB). With an offset from a real meter or a calibrator they read dB SPL, but the result is still not a certified measurement.

Worked example

The first example file is a quiet 1 kHz tone with a short, loud 4 kHz burst after 1.1 seconds. Suppose you know from a calibrator that this recorder maps full scale to 94 dB, so you set the calibration offset to 94 dB. The result:

• LAeq 81.2 dB, LCeq 79.8 dB, LZeq 80.2 dB.
• LAFmax 88.5 dB but LASmax only 82.4 dB.
• LCpeak 91.8 dB.
• LA10 88.1, LA50 84.1, LA90 74.0 dB.

The burst lasts 0.2 s. The Fast meter almost catches it (it reads about 1 dB below a steady tone of the same loudness), while the Slow meter, with its one-second memory, reads about 7 dB lower. LA90, the level exceeded 90 % of the time, is the steady tone underneath: a good estimate of the background. With a calibration the tool adds a comparison: 81 dB(A) is like loud street traffic.

How to read the result

• LAeq is the number most rules and studies use: the A-weighted energy average over the whole file. A-weighting follows how the ear hears quiet sounds and ignores much of the low rumble.
• LCeq minus LAeq tells you how much low-frequency sound there is. In the second example (hum and noise) LCeq is about 17 dB above LAeq: the hum dominates.
• LAFmax / LASmax show short events. Fast follows quick changes; Slow smooths them.
• LCpeak is the highest instantaneous value (impulse noise).
• LA10 / LA50 / LA90 come from the Fast level sampled 100 times a second after the first second.
• The weighting table compares the digital filter with the published curve at the file's sample rate. They agree within 0.3 dB up to 10 kHz at 44.1 and 48 kHz, within about 0.5 dB at 32 kHz.

Tips and common mistakes

• Turn off noise suppression, echo cancellation and automatic gain in the recording app. They can change levels by 10 dB or more.
• A clipping warning means the loudest parts were cut off; the maxima and the peak then read too low. Record at a lower gain.
• Calibrate once per device: record a known level with offset 0 and use the difference as the offset.
• Low sample rates (8 or 16 kHz) hold no high frequencies, so A and C levels miss them. The tool warns you.
• Stereo files are mixed to mono; pick one channel if only one microphone faced the source.

FAQ

Is this a certified sound level meter?

No. The weighting and time constants follow the IEC 61672 formulas, but the microphone, the recorder and their calibration decide the accuracy.

Why does my uncalibrated recording show negative decibels?

Without an offset, 0 dB is a full-scale sine. Real recordings sit below full scale, so they read negative. Add an offset to get dB SPL.

What is the difference between Leq and Lmax?

Leq is the average energy over the whole time; Lmax is the loudest moment after time weighting. A single door slam raises Lmax a lot and Leq only a little.

Which file formats work?

WAV with 8, 16, 24 or 32-bit PCM or 32-bit float, from 8 to 192 kHz and up to 10 minutes. Convert MP3 or M4A to WAV first.