Mastering
True-peak limiting for streaming
Why a mix that peaks at -0.3 dB can still clip a phone, what true-peak detection does about it, and how loudness, balance and the limiter divide the work for streaming delivery.
Hear and inspect the tools in this guide
Loop the same phrase, change one control at a time and compare at similar playback loudness. The linked walkthroughs are product demonstrations, not measurement results for this guide.
Pictured: Finish Him (Limiter) — Choose Transparent, Punch, Dense or Safe limiting.
Watch the Finish Him video walkthrough →
The peak you cannot see
A digital meter shows sample values. Playback rebuilds a continuous wave from those samples, and the rebuilt wave can overshoot between them — sometimes by more than a decibel on bright, loud material. That overshoot is the inter-sample, or true, peak.
It matters because the next stop after your bounce is lossy encoding and a phone DAC, neither of which leaves room for overshoot. The distortion you never heard in the studio arrives in the listener’s kitchen instead.
- Bounce a loud mix at your usual ceiling and measure its true peak against its sample peak.
- Treat -1 dBTP as a working delivery ceiling, not a target to creep past.
- Measure the rendered file, not the session meter. The file is the record.
Draft while you work, full precision when it counts
Careful true-peak detection costs CPU. That is a fair trade while faders are still moving and a poor excuse on the bounce. Work at a setting that keeps the session responsive, then render at the highest precision the limiter offers.
Finish Him names this workflow directly: true-peak detection from Draft to Ultra, with Render Ultra for offline bounces. Draft the decision, render the result. Before bouncing, play the loudest section once at the render setting so nothing surprises you mid-export.
- Track and mix with detection light. Keep the session moving.
- Switch to the render setting before bouncing and audition the loudest chorus.
- Bounce offline. Check loudness and true peak on the file you actually made.
Loudness is a balance that survived, not a number
Streaming normalisation turns loud masters down, which ended the old war and started a quieter one: whose choruses still lift. Two masters at the same integrated loudness can feel completely different, because listeners experience moments — the lift into a chorus, the drop before a hook — not averages.
Protect contrast while you mix. Let verses sit genuinely quieter, automate level before dynamics, and remember that headroom you keep early is loudness you can still spend late, on purpose.
- Mark the three moments the track must feel biggest. Protect those first.
- Compare verse and chorus short-term loudness. Twins mean something is crushing the lift.
- Normalise-check the bounce on streaming-style playback and confirm the lifts survived.
When the limiter complains, believe it
Gain reduction that pumps, cymbals that splash and choruses that feel smaller than the verse are not limiter settings to tune away. They are the limiter reporting a balance problem in the only language it has.
Fix balances first, then limit for level. If reaching your loudness takes more than a few decibels of steady reduction, the mix wants revisiting, not crushing. A limiter’s honest job is the last few decibels, not the whole journey. Encoding and sample-rate conversion can change peaks; the amount depends on the material and conversion. Choose a ceiling appropriate to the destination and measure the exported file; no single value guarantees every codec or playback device.
- Note average gain reduction in the loudest chorus. Steady and deep is a warning.
- Turn the limiter off, rebalance what was fighting it, then re-limit.
- Compare at matched loudness. Better should survive quieter.
Side-by-side comparison of related plug-ins
Compare the Fat Berry DSP plug-ins referenced in this guide by price, category, formats, and primary studio role. Every licence is a perpetual single-user licence (3 machine activations, 14-day full refund policy).
| Plug-in | Category | Regular Price | Formats | Primary Studio Job |
|---|---|---|---|---|
| Finish Him | Limiter | $79.99 USD | VST3, AU | Choose Transparent, Punch, Dense or Safe limiting. |
| End Times | Limiter | $59.99 USD | VST3, AU | A loudness chain: clipping, spectral limiting and a true-peak ceiling in one. |
References and comparison method
This explanation uses the linked product and host references. Suggested listening steps are for your own session; they are not a report of controlled listening or compatibility testing.
Use the actual product controls, compare at similar playback loudness and check the result in the arrangement. Settings and behaviour vary by recording and software release.
Sources and further reading
Questions, answered
What ceiling should I deliver for streaming?
Around -1 dBTP true peak is the working ceiling used throughout this guide. It leaves room for lossy encoding and phone playback. Measure the rendered file with an appropriate true-peak meter; session meters differ and may show sample peaks, true peaks or both.
How loud should a master be in LUFS?
There is no single target that survives every service, and normalisation turns louder masters down anyway. Chase contrast instead: verses that breathe make choruses feel loud at whatever integrated level you deliver.
Why does my master distort on a phone but not in the studio?
Possible causes include reconstructed peaks, encoding, a clipped earlier stage or the playback device. Compare the final file, measure its true peak and trace the actual source of distortion. A limiter ceiling alone cannot diagnose every playback problem.
Draft or Ultra true-peak detection?
Draft while you work, because careful detection costs CPU and the session must stay responsive. Ultra — Render Ultra in Finish Him — on the bounce, when the decision is final and the file is the record.