True peak is a measurement of the actual analog waveform that gets reconstructed between your digital samples, and it can be higher than any single sample value your normal meter shows. Those hidden peaks are called inter-sample peaks, and they are why the standard advice is to set your limiter ceiling to -1 dBTP rather than 0. That 1 dB of headroom protects your master from clipping on consumer playback and from the extra distortion that appears when streaming platforms convert your file to a lossy format. Here is why the sample meter lies to you.
Sample peak vs true peak
Digital audio stores your waveform as a series of discrete samples - snapshots of the signal's level taken tens of thousands of times a second. A normal peak meter reads the value of those individual samples and reports the highest one. That is sample peak. It is easy to measure and it is what most basic meters show.
The problem is that the samples are not the actual sound. When your file is played back, a digital-to-analog converter reconstructs a smooth continuous waveform by drawing curves between the sample points. Those curves can rise higher than any of the individual samples they connect. The real analog level at the crest of one of those curves is the true peak, and it can sit above your highest sample. So a track that reads exactly 0 dBFS on a sample meter can actually reconstruct to +0.5 or +1 dBTP in the real signal - over the ceiling, even though the sample meter swore it was fine.
What inter-sample peaks are
An inter-sample peak is exactly that gap: a moment where the reconstructed waveform between two samples overshoots the level of the samples themselves. They are most common in dense, heavily limited material where a lot of peaks are pushed right up near the ceiling. The harder you limit and the closer you push to 0 dBFS, the more inter-sample overs you create, precisely because you have flattened everything against the top where these overshoots happen.
You cannot see them on a sample-peak meter, because by definition they live between the samples. You need a true-peak meter, which oversamples the signal - reconstructs it at a higher resolution - to estimate the real inter-sample level. The unit it reports is dBTP, decibels true peak. Any true-peak meter or loudness meter with a true-peak mode will show you these overs.
Why inter-sample peaks matter
If they are invisible on a normal meter, why care? Because they cause real, audible problems downstream.
- Clipping on consumer playback. Many consumer DACs and cheap playback devices do not have headroom to reconstruct a peak that overshoots 0 dBFS. When the true peak exceeds the ceiling, that device clips it, adding distortion that was never in your file - and you have no control over which device a listener uses.
- Lossy encoding distortion. This is the big one. When a platform transcodes your WAV into a lossy format like AAC, Ogg Vorbis, Opus or MP3, the encoding process itself can push peaks even higher than they were in your original. A master that was already flirting with 0 dBFS can end up several tenths of a dB over after encoding, producing crackle and distortion on the streamed version that you never heard in your studio.
- Unpredictable across platforms. Each platform uses a different codec at a different bitrate, so the exact overshoot varies. You cannot tune for one, so you leave margin for all of them.
Why -1 dBTP is the standard ceiling
Setting your true-peak ceiling to -1 dBTP gives you a 1 dB buffer for all of that to happen inside without clipping. If encoding pushes your peaks up by a few tenths of a dB, they still land safely below 0 and reconstruct cleanly on any device. It is cheap insurance - 1 dB of level you will never miss, because streaming normalizes loudness anyway and you are not competing on raw peak level.
Some engineers go to -1.5 or even -2 dBTP for very dense, heavily limited masters that create more inter-sample overs, or for material destined for low-bitrate encoding. For most streaming releases, -1 dBTP is the reliable default. The key is that your limiter must be in true-peak mode - a limiter set to a -1 dB ceiling in sample-peak mode will still let inter-sample overs through, because it is only watching the samples. For how the ceiling fits into overall loudness, read how loud a master should be.
How to check and set true peak
The workflow is straightforward once you know what to enable.
- Use a true-peak meter. Put a loudness meter with true-peak mode on your master and watch the dBTP reading, not just the sample-peak number.
- Enable true-peak limiting. In your final limiter, turn on true-peak or inter-sample detection and set the ceiling to -1.0 dBTP.
- Meter the whole song. The highest true peak can occur anywhere - a single snare hit in the bridge - so check the maximum across the entire track.
- Do not master to 0. Even if a sample meter shows exactly 0 dBFS with no overs, your true peak is almost certainly above 0. Always leave the -1 dB margin.
True peak is the ceiling that protects everything you worked on from being wrecked in the last step before a listener hears it. Set it once, at -1 dBTP, in true-peak mode, and your master survives the transcode intact. For the export settings that feed a clean master, read how to export a WAV for mastering, and browse the full mastering guides for the loudness picture around it.
Free download: keep the streaming loudness targets cheat sheet - the LUFS and true peak for every platform - right next to your DAW. PNG and printable PDF.
Try it on your own track. Upload a mix to Sauce Mastering and hear it mastered for streaming - your first master is free, no plugins, no card.