How it works
The server uses FFmpeg loudness analysis to estimate inter-sample peaks. This is distinct from the browser crest meter's sample peak.
Worked example
If the reading is -0.2 dBTP and your ceiling is -1 dBTP, at least 0.8 dB of attenuation is needed before other processing. Re-measure the final encoded file.
Read the result
Stored samples do not show every point in the reconstructed continuous waveform. True-peak analysis estimates inter-sample maxima, which is why dBTP can exceed the highest sample reading. The meter's result is a property of the analyzed file and method, not a listening-quality score.
Common mistakes
Do not relabel sample peak as true peak. Leave codec headroom when required, and re-measure after processing or lossy encoding because the reconstructed peaks can change. Attenuating a clipped source does not remove distortion already printed into it.
Questions musicians ask
What is true peak in mastering?
True peak estimates the highest reconstructed waveform level, including values between stored samples. It is normally expressed as dBTP. MixGlint's server meter uses FFmpeg loudness analysis rather than simply scanning stored sample amplitudes.
What is the difference between sample-peak and true peak?
Sample peak is the largest discrete sample. True peak estimates the reconstruction between samples as well. The browser crest-factor tool reports sample peak; use this page when the delivery requirement specifically names true peak.
How to reduce true peak?
For straightforward headroom, reduce gain and re-measure; for example, moving -0.2 dBTP to -1 dBTP requires at least 0.8 dB attenuation before other changes. A limiter may control peaks differently. Spotify's own published guidance is linked below.
Sources and calculation notes
From measurement to mastering
Understand the measurement, then compare your own mix and reference. Your ears remain the final judge.