AudioMundo

A plain-language reference desk for sound, audio formats, and levels

What Does dBTP Mean? True Peak vs. Sample Peak

dBTP means decibels relative to full scale, true peak. It is a peak measurement taken on the waveform as it would be reconstructed by a converter — including the peaks that fall between samples — rather than on the sample values themselves. The measurement method is defined in ITU-R BS.1770, the same recommendation that defines the loudness algorithm behind LUFS.

Sample peak vs. true peak

A sample peak meter reads the largest absolute sample value in a block and expresses it in dBFS. It answers one question: did any stored sample hit the top of the number range?

A true-peak meter answers a different question: where does the continuous signal peak once the samples are converted back into a waveform? Because a digital signal only stores values at discrete instants, the reconstructed curve passing through those points can rise above the highest stored sample. That excess is called an inter-sample peak. A file can therefore read −0.1 dBFS on a sample-peak meter and still exceed 0 dBTP.

BS.1770 handles this by specifying oversampling before peak detection — the signal is upsampled so the meter can see between the original samples — and the result is reported in dBTP.

Why the suffix matters

A decibel value on its own is a ratio, not a level. The NIST guide to the SI is explicit that a level expressed in decibels must state the reference quantity it is compared against (NIST Guide to the SI, Chapter 8). For dBFS and dBTP that reference is digital full scale, so 0 is the ceiling and working values are negative. This is the same reasoning that separates dBu, dB SPL, and dBFS on a spec sheet — see what a decibel actually measures and how to read a decibel chart.

Where dBTP shows up

  • Delivery specifications. Broadcast and platform specs usually pair a loudness target in LUFS with a separate true-peak ceiling in dBTP. The ATSC recommended practice A/85 covers loudness and true-peak practice for US digital television and publishes a short loudness and true-peak quick reference alongside the document (ATSC A/85). The ceiling it states sits below 0 dBTP rather than at it, precisely to leave room for inter-sample peaks.
  • Metering plugins. A meter labeled "TP" or "dBTP" is running the oversampled measurement; one labeled "peak" or "dBFS" usually is not.
  • Encoder headroom. Lossy encoding does not preserve the waveform sample for sample, so a file that was exactly at the ceiling before encoding can decode slightly above it. Delivery specs set the ceiling below full scale for this reason too, which is part of why the same master behaves differently after platform processing — see loudness normalization on streaming services.

Reading a meter correctly

Three numbers usually appear together, and they are not interchangeable:

  • Integrated loudness (LUFS) — a long-term, perceptual average over the whole program.
  • Short-term or momentary loudness (LUFS) — the same algorithm over a sliding window.
  • True peak (dBTP) — an instantaneous maximum, not a loudness figure at all.

A quiet-sounding mix can have a high true peak, and a loud-sounding one can sit well under the ceiling. The distinction between the perceptual scale and the full-scale digital scale is covered in LUFS vs. dBFS.

Where to look next

The measurement definitions — filter curves, gating, and the oversampling requirement for true peak — live in BS.1770 itself. For US television delivery practice built on top of it, A/85 is the document to read. Both are revised periodically, so check the publisher's page for the current edition rather than a secondhand summary.

Sources