True peak, in one sentence
True peak (measured in dBTP — decibels True Peak) is the highest level the waveform actually reaches once it's reconstructed into a continuous analog signal, including the peaks that fall between digital samples. Sample peak only looks at the stored samples; true peak looks at what a converter or a lossy codec will actually produce. A file can read -0.1 dBFS sample peak yet hit +0.5 dBTP after conversion.
Sample peak vs true peak
Sample peak is the loudest individual sample in the file. True peak is the loudest point of the reconstructed waveform, which can sit between two samples (an inter-sample peak, or ISP). Because digital audio only stores samples, the true peak is estimated by upsampling the signal and finding the maximum of the interpolated curve. This is why a master that looks safe on a normal peak meter can still clip on playback.
Why true peak matters for streaming
Two stages introduce inter-sample overs. First, the digital-to-analog converter in a phone or DAC reconstructs the waveform and can overshoot 0 dBFS, causing audible distortion. Second, and more importantly, lossy codecs (AAC, MP3, Ogg/Opus used by Spotify, Apple Music and YouTube) re-encode the audio and routinely push inter-sample peaks higher. A master clipped to 0 dBFS sample peak will almost always distort after the platform's codec. That's why mastering engineers leave true-peak headroom.
How true peak is measured
The ITU-R BS.1770 standard specifies measuring true peak by oversampling the signal at least 4× and taking the maximum of the interpolated waveform. Higher oversampling (8×) catches worst-case peaks more precisely. Quantara Music measures true peak with a 4× polyphase FIR (Kaiser-windowed) oversampler and keeps a small internal safety margin, so the real inter-sample peak stays at or below the ceiling you choose — the in-browser preview is bit-for-bit identical to the exported file.
What dBTP should you target?
The practical standard is a ceiling of -1 dBTP, which leaves enough headroom to survive codec re-encoding on every major platform. For very loud club or DJ masters some engineers use -0.3 to -1 dBTP; for extra safety on lossy platforms, -2 dBTP is common. On Quantara you set the ceiling and a true-peak-safe limiter enforces it, so you don't have to guess.
How to keep your master true-peak safe
Leave true-peak headroom instead of clipping to 0 dBFS. Use a true-peak limiter (one that oversamples), not a simple sample-peak ceiling. Check the true-peak reading on a BS.1770 meter before export. And remember: making a master louder by removing headroom is self-defeating on streaming, because the platform turns loud masters down on playback anyway — you keep the distortion and lose the loudness.
Frequently asked questions
Sample peak is the loudest stored sample; true peak is the loudest point of the reconstructed analog waveform, including peaks that fall between samples. True peak is measured by oversampling and is always equal to or higher than sample peak.
-1 dBTP is the safe, widely-used standard for streaming. It leaves headroom so lossy codecs (AAC, MP3, Opus) don't push the signal into distortion after re-encoding. -2 dBTP is even safer for loud masters.
Because -0.1 dBFS is a sample-peak reading. Spotify's lossy codec re-encodes the audio and can lift inter-sample peaks above 0 dBFS, causing clipping. Master to a true-peak ceiling (e.g. -1 dBTP), not a sample-peak ceiling.
Yes. Quantara uses a 4× oversampled true-peak limiter with a safety margin, and shows a live dBTP reading, so the exported master stays at or below your chosen ceiling.