Mastering & delivery
True peak vs sample peak: why the meters can disagree
Understand dBFS and dBTP, check reconstructed peaks and verify the actual file after gain changes or conversion.
Sample peak is the highest stored sample value, expressed in dBFS. True peak estimates the highest reconstructed waveform level, including peaks between samples, expressed in dBTP. A file can remain below 0 dBFS on the sample meter while exceeding that level on a true-peak meter. The meters are not meant to report the same thing.
By Spectraforge Audio2 min read
Published
Guide review: Captum Records and AS Mastering. AI-assisted writing and source checks. How these guides are made.
Stored samples and reconstructed peaks answer different questions
Use a meter implementing the appropriate measurement standard and analyse the delivered file. ITU-R BS.1770 defines loudness and true-peak methods; a LUFS result cannot replace a peak result. Neither result, alone, tells you whether distortion is audible.
Worked arithmetic: add room without adding more limiting
Hypothetical example: a clean floating-point render reads −0.3 dBFS sample peak and +0.2 dBTP true peak. Lowering its gain by 0.7 dB would put the measured true peak around −0.5 dBTP if nothing else changes. This illustrates simple attenuation; −0.5 dBTP is not a universal delivery recommendation.
Remeasure after writing the actual destination format. If a fixed-point export already clipped, attenuation cannot restore the lost waveform: return to the unclipped session or higher-precision render. Likewise, turning down a deliberately flattened snare leaves the flattened sound intact. Peak room and restored dynamics are separate jobs.
Check the limiter’s control mode and everything after it
True-peak metering displays overs; true-peak limiting attempts to constrain them. Confirm the control mode is enabled when you need a true-peak ceiling. Oversampling, soft clipping and a control called Guard do not automatically provide that function.
Check gain, faders, EQ and normalisation after the limiter. If the destination specifies a maximum dBTP, measure the entire finished file against it. For a rate-converted or lossy version, inspect that version too: conversion can change peaks. Choose margin for the actual destination instead of treating 0.0 as an objective.
Suggested verification routine
Render the intended master and open it on a clean path. Reset the meter, play or analyse the full file, and record sample peak and maximum true peak with the exact filename. Audition the hard snare, a bright chorus and the final decay. If the delivery includes AAC or MP3, decode that preview and repeat the check.
After reducing drive or changing the ceiling, level-match the musical comparison to the earlier version. Listen for snare weight and bass motion. A compliant number can coexist with an over-limited sound; keep the version that meets the delivery and preserves the performance.
Technical references
- ITU-R BS.1770-5: programme loudness and true-peak measurement
- FabFilter Pro-L 2: output ceiling, metering and true-peak limiting
Listening exercises are proposed practice routines. Worked examples illustrate the stated setup; they are not reports of our listening tests. Editorial process and corrections.