A sample-peak meter can remain below full scale while reconstruction or later encoding produces a higher peak. Compare the intact master with the decoded AAC delivery using a capable true-peak meter before blaming the recording. AAC does not always become louder or distorted; crackling alone does not prove this cause.
Identify what the meter measured
Sample peak checks the stored sample values. True-peak measurement estimates the maximum of the reconstructed waveform, including between samples. ITU-R BS.1770-5, approved in November 2023, describes this distinction and its Annex 2 measurement approach. A meter labeled only peak is not necessarily performing that estimate.
Apple's Digital Masters brief also explains why an encoded file deserves checking separately from its PCM master. Its Apple AAC tools provide one concrete roundtrip workflow, rather than a promise about every codec, operating system or video platform. Treat sample-peak readings in dBFS and true-peak readings in dBTP as different reported quantities.
Locate the first changed stage
Keep a lossless working master and the exact delivered AAC or video file. Select an affected transient, such as a sharp musical accent, with surrounding sound. Record encoder, bitrate, sample rate and the file version being inspected.
- Listen to the master at a comfortable level with other processing bypassed.
- Measure sample peak and true peak using a documented external tool.
- Decode the exact encoded delivery for comparison, preserving floating-point headroom where supported.
- Measure that decoded result and compare the same event at similar listening volume.
- Repeat playback on another supported device to separate file evidence from a device-specific fault.
Decoding into a fixed-range integer file that clips overs immediately can hide their original size. Use a tool that can inspect encoded-file clipping or retain above-full-scale values for the measurement stage. This is a diagnostic copy, not permission to deliver an overloaded file.
A hypothetical export comparison
Imagine a short film whose untouched master sounds clean and whose editor reports a sample peak below zero. Its AAC export roughens one cymbal accent. A roundtrip measurement finds higher true peaks around that accent, while the master and a quieter fresh encode compare cleanly.
That combination supports investigating downstream overload. It is a hypothetical result, not a test of Imagild or a universal AAC behavior. If the original already contains flattened distortion, making a quieter encode cannot recreate the missing waveform. If measurements do not support overload, investigate the microphone, playback buffering, effects or speakers instead.
Correct from the master and recheck
On a copy of the intact mix, reduce gain or revise a suitably configured limiting stage, following the destination's actual peak requirement. Do not copy a music-service ceiling or loudness target into every film contract. Export again from the master with the intended encoder, then repeat the decoded-file check.
Accept the delivery when measured headroom meets the receiving specification and the important accents remain clean in the delivered file. Reject a pass based solely on an editor meter or a quieter listening comparison. Preserve the original mix, measured versions and settings so a later platform transcode can be investigated separately.
Sources
- ITU-R BS.1770-5 (November 2023), Annex 2: True-Peak Measurement
- Apple Digital Masters: Encoded-File and Inter-Sample Clipping Checks
Prepared by Imagild Editorial with AI assistance and reviewed against the linked sources on October 13, 2026. Examples are hypothetical; external procedures do not establish Imagild features or guarantee results.