glossary
K-weighting shapes the signal before LUFS is calculated so the meter behaves more like human hearing.
K-weighting is part of the ITU-R BS.1770 loudness standard. It applies a defined filter curve before calculating loudness, reducing the influence of energy our ears do not perceive as strongly.
This is one reason LUFS is more useful than raw peak level when judging how loud a mix will feel to listeners.
Deep sub energy can eat headroom without raising perceived loudness in the same proportion. K-weighted loudness reflects that more realistically than a simple sample peak meter.
A mix can therefore be peak-heavy, bass-heavy, and still not feel loud. The meter is showing a real translation issue, not a bug.
You can see K-weighting in real numbers: in our scripted cross-check against FFmpeg's EBU R128 loudness measurement, a 100 Hz tone and a 10 kHz tone generated at the same level measured −21.85 LUFS and −16.65 LUFS integrated respectively — the curve discounts the low tone and emphasises the high one.
Meter Core's K-weighted integrated readings track FFmpeg's within 0.04 LU on those filter-curve tones, and within 0.05 LU worst-case across the full 8-case set. Full method and per-case numbers: /meter-core/accuracy/.
You do not need to set K-weighting manually on a proper LUFS meter; it is built into the measurement. What matters is understanding why LUFS and dBFS answer different questions.
Meter Core uses loudness metering alongside peak and stereo checks so one number does not carry the whole decision.