Headroom
The space between your average signal and the digital ceiling — and why you need more of it than you think.
Headroom is the gap between your signal's average level and the maximum the system can handle without distortion. In digital audio, the ceiling is 0 dBFS. Every dB below it is headroom you've banked.
Treating headroom carelessly is the most common cause of rough-sounding mixes. Even when nothing is technically clipping, running everything close to the ceiling stacks problems through the chain.
Why You Need Headroom
For dynamics. Compressors, limiters, and saturators all need headroom to do their job. A signal already squashed against 0 dBFS gives a compressor nothing to work with.
For peak transients. Drum hits, vocal consonants, and percussion peaks are often 6–10 dB louder than the average level of the source. Without headroom, those peaks clip.
For summing. Mixing 20 tracks together adds energy. A mix that averages −6 dBFS per track will sum to a mix that peaks well above 0 dBFS without intervention.
For mastering. A mastering engineer needs headroom to add their own processing. A mix delivered at −0.5 dBFS gives them no room to work.
Headroom by Stage
| Stage | Target peak | Notes | |-------|-------------|-------| | Tracking | −12 to −6 dBFS | Gives compressors and EQ in the chain plenty of room. | | Mixing | Bus peaks at −6 dBFS or below | Master bus peaking around −6 to −3 dBFS gives mastering enough room. | | Mastering input | −6 dBFS peak | The standard recommendation from most mastering engineers. | | Final master output | −1 to −0.5 dBFS true-peak | Below 0 dBFS true-peak after limiting. Avoids issues during lossy encoding. |
True Peak vs Sample Peak
A signal can hit −0.5 dBFS in the DAW but exceed 0 dBFS once decoded for playback. This happens because the analogue waveform between sample points can be higher than any individual sample.
Lossy compression (MP3, AAC, Opus) introduces additional intersample peaks during the encoding process. A safe target for streaming delivery is −1 dBTP (decibels true-peak). FabFilter Pro-L 2, iZotope Ozone, and Waves L2 all have true-peak limiting modes that detect and prevent this.
Gain Staging in Practice
The textbook gain staging philosophy: every stage of your signal chain should operate at its sweet spot, neither too quiet (noise floor problems) nor too hot (distortion and lost headroom).
For tracking through analogue gear modelled by a plugin (UA Apollo, Slate Virtual Mix Rack), the plugin emulation often expects a signal calibrated at +4 dBu reference. That typically means the plugin sounds best with a signal averaging around −18 dBFS at its input.
For DAW-native plugins, the input level matters less mathematically (32-bit float internal precision means no clipping), but a sensible average level (−12 to −18 dBFS) keeps your meters useful and your faders in their effective range.
The Faders Problem
If your tracks are tracked too hot (peaking near 0 dBFS), every fader needs to come down significantly to make room for summing. A fader sitting at −20 dB doesn't have much usable downward range. You've lost fine control.
If your tracks are too quiet (peaking at −40 dBFS), faders need to be pushed up to bring them to a useful level — and the noise floor rises with them.
The sweet spot: tracks recorded with peaks around −6 to −12 dBFS sit naturally near unity (0 dB) on the fader. Small fader moves produce useful changes.
Common Mistakes
"Mixing for the limiter." Pushing every track loud, knowing the master limiter will catch peaks. The limiter does catch them — but at the cost of dynamics that you'll never get back.
Forgetting gain compensation on plugins. A compressor reduces level by 6 dB; that's part of its job. Use the make-up gain or output knob to restore level. Stacking processors that all reduce level produces a quiet mix that needs to be cranked back up at the end, often clipping in the process.
Mastering input too loud. A mix delivered to a mastering engineer with peaks at −0.5 dBFS gives them nothing to work with. They can't add bass or compress the master bus without the mix clipping. Hand off mixes peaking at −6 dBFS (or quieter) to give the next person room.
Believing 0 dBFS is "louder." A mix at peak −1 dBFS isn't quieter than a mix at peak 0 dBFS — your DAW's monitor level controls what you hear. The 1 dB difference at the digital level doesn't change perceived loudness on playback in any meaningful way.