Sample peak measures the loudest individual sample stored in your file, in dBFS. True peak measures the loudest point of the reconstructed analog waveform your file will actually become on playback, in dBTP. That second number is almost always higher, and it’s the one that decides whether your master survives streaming compression intact.
Here’s the working rule: measure and limit with a true-peak tool, and set your ceiling at -1 dBTP as your starting point for streaming delivery. This isn’t arbitrary. It comes straight from the oversampling method defined in ITU-R BS.1770-5, the standard on which most loudness and true-peak meters are built.
- Sample peak reads what’s stored. True peak reads what plays back.
- The gap between the two is typically 0.5 to 1.5 dB on bright, transient-heavy material.
- That gap is exactly the headroom a -1 dBTP ceiling is built to absorb.
Key Takeaways
True peak, not sample peak, sets the real safety margin for streaming and broadcast delivery, and a -1 dBTP ceiling is the reliable starting point for most releases.
| Point | Details |
|---|---|
| Two different measurements | Sample peak reads stored dBFS values; true peak reads the reconstructed dBTP waveform after playback. |
| Expect a real gap | Inter-sample peaks commonly add 0.5 to 1.5 dB above sample peak on bright, transient material. |
| Set your ceiling by target | Use -1.0 dBTP for streaming, -2.0 dBTP for low-bitrate delivery, and check broadcast specs separately. |
| Limiter goes last | Never add EQ, saturation, or sample-rate conversion after your final true-peak limiter. |
| Verify the exported file | Measure the exported WAV and its re-encoded lossy version, not just the in-session reading. |
Table of Contents
- True Peak vs Peak: The Signal Math Behind the Gap
- Why True Peak Matters for Codec Delivery and Playback
- How Do You Measure True Peak on a Finished Master?
- How to Set a True-Peak Limiter Ceiling
- A Practical True-Peak Checklist Before You Deliver
- Sources
- FAQ
True Peak vs Peak: The Signal Math Behind the Gap
A sample-peak meter checks the height of each stored number and reports the tallest one. That’s it. It has no idea what happens between samples when a digital-to-analog converter reconstructs the continuous waveform those samples represent.
That gap between samples is where inter-sample peaks (ISPs) live. Digital audio isn’t a staircase, it’s a series of snapshots that get smoothed back into a continuous curve on playback. When two adjacent samples sit close to full scale but the true waveform between them arcs higher, the reconstructed peak overshoots what either sample shows. A kick drum transient or a cymbal crash, both rich in high-frequency content, is a classic trigger.
True-peak meters catch this by oversampling: they interpolate extra data points between existing samples to approximate the reconstructed curve before measuring the maximum. ITU-R BS.1770-5 sets a baseline of 4x oversampling, effectively upsampling 48 kHz audio to 192 kHz for measurement purposes. Higher oversampling ratios exist, but 4x already catches the overwhelming majority of real-world overshoot.
| Metric | What it measures | Unit |
|---|---|---|
| Sample peak | Highest stored sample value | dBFS |
| True peak | Highest reconstructed waveform value | dBTP |
| Typical ISP gap | Difference between the two on transient material | 0.5 to 1.5 dB |
Most program material stays inside that 0.5 to 1.5 dB range. Very bright, heavily limited, or aggressively saturated masters can exceed it, which is exactly the material where true-peak checking earns its keep.
Why True Peak Matters for Codec Delivery and Playback
A master that reads 0 dBFS in your DAW can clip after it leaves your session, and it usually happens in one of two places.
- Lossy encoding. Converting to AAC or MP3 involves its own reconstruction and filtering, and that process can push transient peaks above the original ceiling. Overshoot gets worse as bitrate drops, which is why a 128 kbps stream is more vulnerable than a 320 kbps one.
- Playback hardware. Consumer DACs perform their own reconstruction on the way to analog. If your file’s true peak sits at 0 dBTP, the DAC has zero margin, and ISPs translate directly into audible distortion or clipping, even though your session meters never showed a problem.
Run the arithmetic: a file mastered to 0 dBFS sample peak with a 1 dB inter-sample peak already sits at +1 dBTP before it ever touches an encoder. Add encoder overshoot on top of that and you’re shipping a file that clips on nearly every playback chain that touches it.
True peak also has nothing to do with how loud a track sounds. Integrated loudness (LUFS) measures perceived loudness over time; true peak measures an instantaneous ceiling. A track can sit at a perfectly reasonable -9 LUFS and still clip, because one transient decided the true-peak outcome regardless of everything around it.
How Do You Measure True Peak on a Finished Master?
Confirming a master meets its ceiling takes a few concrete checks, not a single meter reading in your DAW.
- Use a meter that’s explicitly BS.1770-compliant and confirm its oversampling setting is at least 4x.
- Measure right after your final limiter, inside the session, as a first pass.
- Export the file, then measure the exported WAV or FLAC separately rather than trusting the in-session reading alone. Dithering and any sample-rate conversion during export can shift peak relationships.
- Re-encode that export to your target lossy format (AAC or MP3) and measure again to catch encoder overshoot directly.
- Compare sample peak against true peak on the same file. The size of that gap tells you how much ISP margin you actually have.
Pro Tip: Keep your true-peak limiter’s oversampling setting locked before you start comparing files. Switching it mid-session gives you inconsistent readings that look like a mixing problem when it’s really a measurement inconsistency.
How to Set a True-Peak Limiter Ceiling
Fixing true-peak overshoot happens in the limiter, not by chasing gain everywhere else in the chain.
- Choose a true-peak-capable limiter with built-in oversampling. A limiter operating on non-oversampled sample-peak alone will let ISPs pass right through even when its own meter reads clean.
- Set your ceiling by delivery target. Start at -1.0 dBTP for general streaming. Drop to -2.0 dBTP for low-bitrate distribution, where encoder overshoot is more aggressive. Some broadcast specs call for -3.0 dBTP, so check the destination’s requirements before finalizing.
- Place the limiter last. Nothing goes after it, no EQ, no saturation, no sample-rate conversion. Any processing downstream of your final limiter can reintroduce peaks you already controlled.
- If overshoot persists, don’t just push the ceiling lower and call it fixed. Pull back upstream gain, tame bright transients with a transient shaper or a gentle high-shelf cut, then re-export and re-measure.
Oversampling ratio matters here too. Pushing to 8x or 16x squeezes out marginal extra accuracy on the harshest transient content, at the cost of noticeably higher CPU load during real-time limiting. For very loud, heavily limited masters, running slightly more headroom than -1 dBTP is worth considering, since aggressive limiting itself tends to generate more inter-sample peaks.
A Practical True-Peak Checklist Before You Deliver
Run this sequence before any file leaves your hands: measure the exported WAV’s true peak, re-encode to AAC or MP3 and measure again, confirm your limiter’s ceiling sits at -1 dBTP (or lower for low-bitrate or broadcast targets), then export once more and verify nothing shifted. Four checks, five minutes, and it catches the overwhelming majority of delivery problems before a listener ever hears them.
Some material doesn’t forgive shortcuts here. Dense, transient-rich mixes, strict platform specs, or a mix you tracked in a room you don’t fully trust acoustically are exactly the situations where a second set of ears matters. That’s the gap outsourcing to a mastering engineer closes. Twisby Records has run this exact verification workflow across 35-plus years of mastering work and holds Apple Digital Masters certification, which means every delivered file has already cleared the true-peak and encoding checks this article walks through. For a deeper look at the full mastering pass beyond true-peak control, see what audio mastering actually covers.
— Kreg
Sources
- Recommendation ITU-R BS.1770-5: Algorithms to measure audio programme loudness and true-peak audio level
- True Peak, dBTP and the Inter-Sample Peak Problem
- What is True Peak? Inter-sample peaks, limiter ceiling and codec clipping | Aumixys
FAQ
Is True Peak Too High if It Reads Above -1 dBTP?
Above -1 dBTP isn’t automatically a failure, but it removes the margin most streaming platforms expect and raises the risk of encoder overshoot clipping the file. For low-bitrate distribution, staying closer to -2 dBTP is safer.
Should I Use True Peak Limiting on Every Master?
Yes. A standard sample-peak limiter can’t detect inter-sample peaks, so a true-peak limiter with built-in oversampling is the only way to catch overshoot a normal meter misses.
What Does True Peak Mean?
True peak is the highest level of the continuous analog waveform your file becomes after digital-to-analog reconstruction, measured in dBTP under the ITU-R BS.1770-5 standard. It’s distinct from sample peak, which only reads the stored digital values in dBFS.
What Should My True Peak Be When Mastering?
Target -1.0 dBTP as a general streaming ceiling, and drop to -2.0 dBTP if you’re delivering for low-bitrate playback or uncertain encoding paths. Some broadcast specifications call for -3.0 dBTP, so confirm the destination’s requirement before finalizing your export.
