Guide

True peak, dither, and what a game-ready SFX file looks like

Updated July 26, 2026

A sound effect can measure −0.1 dBFS in your DAW, pass every meter you own, and still crackle on a player's laptop. This is the short version of why, and the delivery spec every Neon Audio pack is shipped against.

Sample peak is not the peak

Your DAW's meter usually shows sample peak: the highest value of the actual samples in the file. That is a real measurement, but it is not the loudest the audio will ever get.

Digital audio is a set of points. The waveform your listener eventually hears is the smooth curve reconstructed between those points by a DAC — or by a sample-rate converter, or by an engine mixing your effect against everything else. That reconstructed curve can overshoot the highest sample. Those overshoots are called intersample peaks, and a file sitting at −0.1 dBFS can reconstruct meaningfully above 0.

True peak is the measurement that catches this. It is defined in ITU-R BS.1770 — the same family of specifications behind EBU R128 loudness metering — and it works by oversampling the signal (4× minimum) to estimate the real analogue peak rather than just reading the sample values. It is reported in dBTP, decibels true peak.

Why this bites harder in games than in music

A mastered song is played back roughly as delivered. A game sound effect is not. Before it reaches a speaker it may be:

• Resampled — your 44.1 kHz file playing through a 48 kHz engine mix.
• Compressed to a lossy codec — Vorbis, MP3 and AAC do not preserve peak levels, and typically overshoot the original.
• Pitch-shifted at runtime — the near-universal trick for making repeated footsteps and gunshots sound less repetitive.
• Summed with everything else playing at that moment.

Every one of those can push a file that was "just under" over the line. A file mastered flat to 0 dBFS has no room to survive any of them.