What a true peak is
Your audio is stored as samples, but a speaker never plays samples. On playback, the converter draws a smooth wave through them, and that wave can rise higher between two samples than either sample did. The highest point of that rebuilt wave is the true peak.
So a master whose samples all sit below 0 dB can still go over on playback. When it does, the converter clips and you hear distortion. A clipper on its own only sees the samples, so it can't catch this. If that is what brought you here, it measured under the ceiling but still distorted is the quick fix and a check to run before you export.
What the TP safety limiter does
Turn on the TP safety limiter switch and HardStop keeps the rebuilt wave under your ceiling, not just the samples. The clipper does most of the work, then a gentle true-peak limiter catches the overshoots the clipper missed. It uses smooth gain reduction, so it doesn't pump.
- It is off by default, and it is in HardStop Basic.
- The Ceiling switches to dBTP (decibels, true peak) while the TP safety limiter is on, and so does the PEAK readout in Pro. The output meters always read true peak.
- The waveform shows gain reduction from both stages. Red is the clipper, green is the TP limiter.
- HardStop measures true peak at 4x oversampling, as ITU-R BS.1770-5 asks, then refines each peak between samples.
- Two settings can still take you over the ceiling: turning Out Gain above 0 dB, and in Pro, setting Mix below 100%. Leave them there when you need the ceiling to hold.
Set your ceiling first
The Ceiling knob starts at 0 dB, so set it before you do anything else, and you won't have to remember later:
- -1 dBTP as your starting point. It leaves a safe margin.
- -0.2 dBTP if you are chasing a competitive level.
That margin also helps later on. Streaming services convert your master to formats like MP3 or AAC, and that conversion can add peaks of its own that no plugin can control. -1 dBTP leaves room for them. Why your songs aren't loud on streaming platforms covers this in more detail.
TP safety limiter on or off?
It depends on how loud you want the master, and how much risk you'll take for it.
| Goal | TP safety limiter | Ceiling | What you get |
|---|---|---|---|
| Loud and safe | On | -1 dBTP, or -0.2 dBTP if competitive | A loud master that likely won't clip the converter on playback. |
| As loud and aggressive as possible | Off | -0.2 dBFS | The most impact, with the risk of some converter clipping on playback. On aggressive material that can add to the feel. |
Neither is wrong. Bounce both, listen on a few systems, and trust your ears. This is where you can actually hear converter clipping distortion on consumer grade playback systems.
Leave the TP safety limiter off on individual tracks too, where more processing comes after HardStop, and while you are recording (see latency below). If another limiter follows HardStop on your master, put HardStop before it and let that limiter handle the ceiling. HardStop's own TP limiter can do the whole job when you need it to.
Latency
- Normal with the TP safety limiter off: zero latency.
- HQ or Ultra: 114 samples.
- TP safety limiter on: 482 samples in total at 48 kHz, in any quality tier.
That amount stays the same however you set the Lookahead knob, and HardStop reports it to your DAW, which compensates automatically. It is still too much to play through, so leave the TP safety limiter off while tracking.
Lookahead and Release (Pro)
In Basic, the TP limiter sets its own lookahead and release. HardStop Pro lets you set both:
- Lookahead is how far ahead the limiter looks for a peak. More lookahead can reduce audible distortion on the limiting stage.
- Release is how fast the limiter lets go after a peak. Longer settings can add groove to the track. Shorter is more aggressive and loud.
Loudness is a separate question
Streaming services turn tracks down based on loudness (LUFS), not peaks. The TP safety limiter keeps your peaks under the ceiling, but it doesn't change how loud your track plays on Spotify or anywhere else. The average top-charting track based on an iZotope blog article from 2025 sits around -8.4 LUFS, far above the -14 target everyone quotes. The post above explains why, and so do these videos:
- This is WHY your songs aren't LOUD on streaming platforms
- I tested Spotify's loudness penalty
- Clipping vs limiting: most mixers get this wrong
Output still going over your ceiling? Check that the TP safety limiter is on, then run the pre-export check in it measured under the ceiling but still distorted.