Read LUFS and true peak on a metal master

One number cannot describe the whole master

A peak meter and a loudness meter answer different questions. Sample peak reports the highest stored sample value. Integrated loudness estimates the average loudness of a defined programme over time. True peak estimates the highest level of the waveform between samples as well as at the samples. For a dense metal mix, these readings can move independently: a short snare transient can set a peak while the guitars and bass shape the longer-term loudness.

ITU-R BS.1770 defines an objective loudness algorithm and a separate method for estimating true peak. The loudness calculation applies frequency weighting and combines channel measurements; the standard also uses gating so very quiet passages do not simply pull the whole reading downward. The recommendation describes an objective estimate, not a perfect prediction of every listener’s response. Playback level, speakers, attention and musical material still matter. ^[1]

LUFS expresses loudness relative to digital full scale. A higher, less negative reading indicates a louder measured programme: for example, −9 LUFS is louder by this measure than −13 LUFS. The difference is four loudness units, but that alone does not tell you whether either master sounds better, clearer or more forceful.

Integrated loudness is a comparison tool, not a sound-quality score

Integrated loudness covers the interval between the meter’s start and stop points. If you measure only a chorus, you are measuring that excerpt; if you measure the complete track, you are measuring the track. Record which interval you used. Otherwise two readings may appear to disagree simply because one includes an intro, breakdown, silence or long fade and the other does not.

Loudness normalization and dynamic processing are also different operations. Normalization changes playback or file gain to meet a level objective. Compression or limiting changes the signal’s dynamics. A service may turn a loud file down during playback, but that does not restore transients or dynamics already changed in mastering. The EBU describes programme loudness, loudness range and maximum true peak as complementary measurements, and notes that suitable dynamics depend on genre, audience and distribution. ^{[2] [3]}

EBU R 128 recommends −23 LUFS for broadcast programmes. That is a broadcast recommendation with a defined context, not a target to impose on every metal album or every music platform. Follow the receiving label, distributor or platform’s current delivery specification when one applies. If none is supplied, document what you measured and why; do not turn a broadcast number into an invented universal rule.

Why true peak can exceed the sample peak

Digital audio stores samples, but playback reconstructs a continuous waveform between them. That reconstructed curve can rise above the largest stored sample. ITU’s true-peak method estimates this inter-sample maximum through oversampling and filtering. A sample peak below 0 dBFS therefore does not guarantee that every later conversion or playback stage will remain below full scale. ^[1]

True peak is commonly shown in dBTP. It is a warning about headroom and possible overload through later processing, not a measure of average loudness. The EBU’s production guidance gives −1 dBTP as its maximum for generic linear PCM production, while also noting that distribution chains and codecs may need different limits. Treat that as a context-specific broadcast production reference, not an automatic mastering ceiling for every commercial release. Check the actual recipient requirements. ^[3]

A limiter can reduce peaks, but lowering a true-peak ceiling may change transient shape and the overall feel if the limiter is doing substantial work. A number that passes a meter is not proof that the snare attack, bass definition or intended contrast survived. Listen to the loudest section and the quieter transition after each technical adjustment.

A controlled meter check before delivery

Use integrated loudness for a defined programme interval, true peak for inter-sample headroom, and listening for musical balance and impact. Save the file version and delivery rule with the measurements.

Next compare the reading with the recipient’s written specification. If the loudness differs from a requested target, decide whether the instruction calls for metadata or an audio-level change; do not assume those are interchangeable. If true peak exceeds the required ceiling, inspect the loudest passages and choose a transparent correction. Re-measure after exporting because conversion, resampling or encoding can alter peaks. Preserve the approved version and record the final filename, measurement interval and results alongside the handoff.

For a simple made-up example, imagine a track measures −10.5 LUFS integrated and −0.4 dBTP. Those figures tell you its measured average level and estimated maximum waveform peak. They do not tell you whether it is too loud for a particular service, whether its limiting is audible, or whether it should be changed. The next step depends on the recipient’s actual delivery rule and a level-matched listening check, not on a comparison with a fashionable number.

A useful note might read: “Approved album master; full track measured from first sample to end of fade; integrated loudness and maximum true peak recorded with the named meter; checked against the distributor specification dated [date].” Replace the bracket with the real instruction and date. That creates a traceable handoff without claiming that one loudness value defines a good metal master.

Sources and the limits of this exercise

The technical definitions here follow the current ITU-R measurement recommendation and the EBU’s broadcast production guidance. The numerical track example is hypothetical and illustrates interpretation only; it is not a measured release, a platform specification or a claim about an artist’s master. Consult the current recipient instructions for an actual delivery. ^{[1] [2] [3]}