Reply to thread

Message: [QUOTE="Shamiso, post: 92784, member: 160"] True peak limiting controls reconstructed waveform peaks that can rise above the highest stored sample values in a digital audio file. A normal sample meter can miss those in-between peaks because it only sees the samples themselves, not the reconstructed waveform they represent. What true peak limiting does is fairly specific. It estimates where the waveform can peak between samples, then makes the limiter react to those estimated peaks instead of only the stored sample values. A true peak limiter is still a limiter, but the detector is watching a more demanding version of the signal. The [B][URL='https://www.aes.org/publications/standards/preview.cfm?ID=54']accurate true-peak metering[/URL][/B] problem is well documented because sample-peak meters can under-read the actual reconstructed maximum. Oversampling and interpolation are used to estimate those missed peaks more closely. [HEADING=2]Detection and limiting are not the same job[/HEADING] How true peak limiting works gets easier to understand when detection is separated from gain reduction. The detector can oversample or interpolate the signal to estimate inter-sample peaks. The limiter then decides how much gain reduction to apply so those estimated peaks stay under its output ceiling. A true peak limiter versus a conventional limiter can therefore produce different gain reduction from the same input. The conventional limiter may see a sample peak that looks safe, while the true-peak detector sees a reconstructed peak above the ceiling. Enabling the true-peak mode can make the limiter work a little harder even when the sample meter barely changes. True peak limiting and oversampling are related, but they are not automatically the same thing. Oversampling can be used inside the detector simply to estimate the waveform more accurately. A plug-in can also oversample other processing stages for different reasons, so seeing an oversampling switch does not prove the limiter is controlling true peaks in the same way. Limiter true-peak detection also does not mean every implementation behaves identically. Filter design, interpolation accuracy, lookahead, release behavior, and the limiting algorithm itself can change the result. Two plug-ins can agree that an inter-sample peak exists and still react differently to it. [HEADING=2]The on-or-off choice depends on the job[/HEADING] The true peak limiting on-or-off decision is mostly about what comes after the master and how much peak control the delivery path needs. If the file will be encoded, resampled, normalized, or otherwise processed downstream, reconstructed peaks matter more than they do in a closed playback chain that you fully control. Knowing when to use true peak limiting also means knowing what it cannot guarantee. A file can leave your limiter under its ceiling and later change after sample-rate conversion or lossy encoding. The limiter controlled the signal it received at that point in the chain, not every version that might be generated later. Whether true peak limiting is necessary therefore has no universal yes-or-no answer. It is useful when you need tighter control over reconstructed peaks, especially near full scale. It is less meaningful as a ritual setting if the next stage will substantially alter the signal and nobody checks the actual deliverable afterward. The same logic applies to true peak limiting on the master. If another limiter, clipper, sample-rate converter, codec, or gain stage follows it, the old ceiling can become stale. A final rendered-file meter pass tells you more than trusting a number from earlier in the chain. [HEADING=2]More peak control can change the sound[/HEADING] Whether true peak limiting sounds bad depends on how much extra work the limiter has to do. If the reconstructed peaks sit comfortably below the ceiling, enabling true-peak detection may change very little. If the master is already pushed hard, catching additional peaks can increase gain reduction around short transients. Extra control can soften attacks or shift the feel of drums when the limiter reacts more often. It can also be practically inaudible on material with more headroom. The useful comparison is level-matched, because a slightly louder version can win a quick A/B even when its transient shape is worse. The true meaning of true peak limiting is not “make the master safer at any cost.” It is better understood as controlling estimated reconstructed peaks at a chosen point in the chain. The [B][URL='https://goldmidi.com/community/resources/what-is-the-difference-between-dbfs-and-dbtp.70/']way dBFS and dBTP describe different peak behavior[/URL][/B] is the reason the extra detector exists in the first place. A limiter with true-peak detection is most useful when you treat it as one part of delivery control rather than a magic compliance switch. Set it because the destination or workflow gives you a reason, listen for any extra transient damage, and meter the finished file that will actually be delivered. True peak limiting works best when the number and the audible result both survive the same final render. [/QUOTE]

Name