True Peak vs Sample Peak: What the Meter Can Miss

Two meters, two questions

A sample-peak meter reports the highest value among the digital samples stored in a file or moving through a digital signal path. It answers a direct question: how close did one of those measured sample values come to full scale?

A true-peak meter estimates a different point in the chain. A digital-to-analog converter reconstructs a continuous waveform from the samples, and the curve between two samples can rise higher than either neighboring value. That possible between-sample maximum is called an inter-sample peak.

The Audio Engineering Society's peak-metering guide explains why a waveform can exceed its nearest sample values and why true-peak measurement estimates those inter-sample excursions. The reading is useful information, not proof that a mix will sound good or that a file meets every distributor's delivery rules.

Sample peak: the stored values

Digital audio represents a signal as a sequence of sample values. A conventional sample-peak meter checks those values and shows the highest one during the measurement period. In many tools the result is displayed in dBFS, decibels relative to digital full scale.

That display is easy to interpret within its scope. It tells you how close the sampled values are to the maximum the digital system can represent. It does not, by itself, describe perceived loudness, musical balance, or every peak that may appear after conversion.

When a sample reaches or exceeds the system's representable limit, clipping can occur. The AES guide describes digital clipping as truncation when samples exceed the maximum or minimum limit. A meter can help reveal the risk, but the sound and the rendered file still need attention.

True peak: an estimate between samples

The reconstructed waveform is not a staircase that jumps only from one stored sample to the next. Between those points, the signal follows a continuous curve. Under some conditions that curve reaches a higher level than the individual samples suggest.

True-peak measurement estimates the curve's possible maximum. The ITU-R Recommendation BS.1770-5 specifies algorithms for measuring audio-programme loudness and true-peak audio level. It provides a measurement method; it does not require every independent release to target one identical peak value.

True-peak readings are commonly expressed in dBTP, while sample-peak readings are commonly expressed in dBFS. The labels help distinguish the measurement basis. A positive or negative number should always be read in the context of the meter, file format, and delivery specification being used.

Why the readings can diverge

The gap between sample peak and true peak is not a fixed offset. It depends on the waveform's shape and phase around the sample points, among other signal conditions. A sharply changing or high-level signal can produce a reconstructed peak that a sample-only meter does not show.

That is why a file can appear to have headroom by sample peak yet produce a higher estimated true peak. The difference does not mean the mix is automatically broken. It means the sample-peak measurement alone may not answer a question about the reconstructed signal.

Encoding and later playback can add additional variables. A true-peak meter helps identify one category of risk before handoff, but it cannot guarantee the same result through every codec, converter, or playback chain.

When each reading is useful

Use sample peak when you need to inspect the digital values in a file or signal path. It is a direct check for how close those values come to full scale and can help locate obvious overs.

Use true peak when a mastering brief, broadcast requirement, distribution specification, or quality-control process asks for an estimate of reconstructed peaks. The current delivery document is the authority for the target and measurement method. Do not substitute an old preset or a number remembered from another platform.

For a music release, compare the requested requirement with the actual exported file. If the brief specifies both sample peak and true peak, check both. If it only names one, do not silently assume it meant the other.

A practical check before delivery

First, confirm which file is being measured. A limiter output meter may not describe a later bounce if the export settings, sample-rate conversion, dither, or processing differ. Measure the rendered deliverable, not only the session's live master bus.

Second, reset the meter and scan the full file. Note the highest sample-peak and true-peak readings, the meter mode, and the file version. If the file is revised, repeat the check on the new render rather than carrying forward yesterday's reading.

Third, listen. A number does not reveal whether a transient feels pinched, a vocal is masked, or a tonal balance is tiring. Metering and listening answer different questions, and neither should be asked to do the other's job.

The useful distinction

Sample peak tells you what the stored sample values reached. True peak estimates how high the reconstructed waveform may rise between them. A producer checking a release can use both readings to understand the file more completely, while following the actual delivery specification instead of chasing a universal rule.

For a broader production perspective, see our guide to mastering for streaming and the distinction between mixing and mastering.