The difference in true peak vs peak is one of the most misunderstood concepts in mastering, and getting it wrong is why a master that looked clean on your meters comes back distorted after it hits Spotify or Apple Music. In short, a normal “peak” (more precisely a sample peak) measures the level of the individual digital samples in your file, while a true peak estimates the level of the actual analog waveform that sits between those samples once your track is converted back to sound. Those two numbers are not the same, and the gap between them is exactly where inter-sample clipping hides. This guide explains why, and gives you the ceiling numbers that keep every release clean.
What a sample peak actually measures
Digital audio is a series of snapshots. At a 44.1 kHz sample rate, your DAW stores 44,100 amplitude values per second per channel. A standard peak meter simply reports the highest of those stored sample values. If the loudest sample in your file sits at -0.1 dBFS, a sample-peak meter says -0.1 dBFS and shows no clipping, because no single stored value went over 0.
The problem is that music is not made of dots; it is a continuous waveform. When a digital-to-analog converter, or a lossy codec like AAC or MP3, reconstructs the smooth curve that passes through those sample points, the real waveform can rise higher than any single stored sample. Picture two samples that sit just under 0 dBFS with a steep transient between them: the reconstructed curve arcs above them both and overshoots 0. Your sample meter never saw it, but the speaker cone and the encoder did.
True peak vs peak: where the extra level comes from
A true-peak meter solves this by oversampling. It internally upsamples the signal, usually 4x, and interpolates the values between your real samples to reconstruct the analog waveform’s actual maximum. That reconstructed maximum is the true peak, measured in dBTP (decibels true peak). Because it captures those inter-sample overshoots, the true-peak reading is almost always higher than the sample-peak reading, sometimes by 0.5 to over 1 dB on dense, heavily limited masters.
- Sample peak (dBFS): the highest stored digital sample. Free, instant, but blind to what happens between samples.
- True peak (dBTP): the estimated highest point of the reconstructed analog signal, found by oversampling. This is the number that predicts real-world clipping.
- The gap: the hotter and more limited your master, the bigger the difference. Sparse, dynamic material barely differs; a wall-of-sound EDM master can show 1 dB or more of hidden headroom loss.
This is why a master can read exactly 0.0 dBFS on a sample meter, look perfectly safe, and still produce audible distortion after streaming compression. The codec re-encodes the waveform, the inter-sample peaks push past 0, and the decoder clips them. You never clipped in your DAW, but the listener hears it.
The ceiling numbers that keep masters clean
Because true peaks can exceed sample peaks, you cannot master to 0 dBFS and expect clean playback. The industry-standard safety margin is a true-peak ceiling of -1 dBTP. Setting your limiter or maximizer to a true-peak output ceiling of -1 dBTP gives roughly a decibel of buffer so that inter-sample overshoots and codec conversion still land below full scale.
- Streaming (Spotify, Apple Music, YouTube, Tidal): keep true peak at or below -1 dBTP. Some platforms recommend this explicitly to avoid distortion after their normalization and encoding.
- Lossy-heavy delivery or podcasts: a slightly safer -1.5 to -2 dBTP gives extra margin for aggressive re-encoding.
- Always enable true-peak limiting. Use a limiter’s true-peak or oversampling mode rather than trusting its sample-peak ceiling. Modern limiters do this internally when you switch the mode on.
To set this correctly, choose a limiter with a genuine true-peak mode, such as FabFilter Pro-L 2, and set its output ceiling to -1.0 dBTP with true-peak limiting engaged. To verify the result independently, meter the finished file with a dedicated true-peak meter like Youlean Loudness Meter, which shows both sample peak and true peak side by side so you can see the gap for yourself.
How to check true peak on your own masters
The workflow is simple once you know what you are looking for. Solo the final master, play the whole track (not just the loudest section, because the true peak might live in a single cymbal hit or vocal consonant), and watch the true-peak readout. If it exceeds -1 dBTP anywhere, lower your limiter ceiling until the maximum true peak sits at or under -1 dBTP.
It also helps to understand why the gap grew historically. During the loudness war, engineers pushed masters harder and harder into the ceiling, and the more you limit a signal the flatter its wave tops become and the more inter-sample overshoots those steep edges create. A gentle, dynamic master with peaks well below the ceiling barely shows any true-peak-versus-sample-peak difference at all, while a brick-walled master routinely hides half a decibel or more of true peak above what its sample meter reports. In other words, the harder you chase loudness, the more true-peak headroom you quietly give away, which is one more reason to master toward a sensible loudness target rather than slamming everything into the limiter.
Do not confuse this with loudness. True peak controls clipping; integrated LUFS controls how loud the track sounds after normalization. A master should satisfy both: integrated loudness around -14 to -9 LUFS depending on genre, and true peak no higher than -1 dBTP. If you are new to setting both at once, our mixing and mastering guide walks through the full chain, and the meters in our free audio tools hub let you read true peak without buying anything. Once you understand true peak vs peak, you will never ship a master that clips on someone else’s speakers again.
