True Peak Limiting Explained: ISP, Codecs, and Streaming Delivery

Sample Peak vs True Peak

Digital audio stores discrete samples. A sample-peak meter reads those values and can show 0 dBFS while the reconstructed analog-like waveform still overshoots between samples. Those overshoots are inter-sample peaks (ISPs). True peak meters estimate peaks of an oversampled reconstruction, reported as dBTP.^[1]

Why care? Lossy codecs (AAC, Ogg Vorbis, etc.) and consumer DACs can exaggerate overs, causing audible clicks or dulling if the encoder’s own limiting kicks in unpredictably. Streaming platforms document loudness normalization and recommend true-peak headroom; Spotify’s artist support explains loudness normalization targets for delivery.^[2]

True peak limiting uses oversampling inside the limiter so gain reduction accounts for those inter-sample estimates—not only sample peaks.

Streaming Targets and Ceilings

Loudness and true peak are related but not identical. You can be −14 LUFS integrated and still spike true peaks if short transients are hot. Conversely, a quiet master can still clip true peak if overs are present. Meter both.

How to Set a True Peak Limiter

If you need more loudness than the limiter can give without crushing, fix the mix: control low-end build-up, tame resonances, and balance sections. Limiters should finish, not rescue chaotic dynamics.

Oversampling, Latency, and Quality

True peak modes usually oversample (2×–8×+). Higher quality can mean higher latency—fine for mastering offline, annoying for tracking through the master chain. Bypass heavy master limiters while recording.

Some free limiters (e.g., LoudMax ISP mode) expose inter-sample options clearly; others only sample-peak. Read the manual. A true-peak meter (Youlean, integrated DAW meters, etc.) should verify the bounce regardless of which limiter you trust.

Creative Exceptions and Genre Notes

Hyper-loud electronic masters still benefit from true-peak awareness even when aesthetic distortion is intentional—clippers before the limiter are a controlled choice. Acoustic and jazz releases often prioritize dynamics; true-peak ceiling still prevents accidental overs without forcing competitive loudness wars.

YouTube, podcasts, and social exports may use different loudness aims than Spotify. Maintain separate export presets: streaming album master, club master, content master. Naming files clearly prevents uploading the wrong version.

Pre-Distribution Checklist

• Meters Integrated LUFS, true peak dBTP, and a quick LRA/dynamic check.
• File Correct sample rate/bit depth per distributor; no clipped render.
• Encode test Listen on phone + earbuds after a lossy encode.
• Docs Note limiter settings in session for recalls and vinyl/other variants.

When stakes are high, re-listen the next day. Ear fatigue hides harsh ISP-related distortion on cymbals and esses. If a distributor QC report flags true peak overs, fix and re-upload—the encode path is what listeners hear.

Mix-First Fixes When the Limiter Protests

If you need more than about 3–4 dB of gain reduction to hit a loudness goal, inspect the mix before blaming the limiter. Common culprits: unfiltered rumble under 808s, resonant snares, unmanaged contrast between quiet verses and smashed drops, and parallel buses that sum hotter than expected.

Practical repairs: clip-gain the loudest phrase peaks, dynamic-EQ harsh bands, shorten reverb tails that pile into choruses, and mono-check low end so the limiter is not fighting stereo sub chaos. A calmer mix lets true peak limiting stay invisible—which streaming listeners reward even when they cannot name the technique.

Keep two masters when needed: a streaming-oriented master with true peak ceiling discipline, and a more dynamic master for other contexts. Name files with `_stream` and `_dynamic` so the wrong version never ships during release week.

Practical Metering Habits

Calibrate a monitoring level and stick to it when judging limiter damage. Louder always wins false A/Bs. Use a true-peak meter on the bounced file, not only the live insert, because render paths and dither can differ slightly from realtime.

If short-term LUFS swings wildly between sections, the limiter may be doing too much musical work. Ride clip gain or bus compression earlier so the final true peak stage only catches peaks. Document ceilings per deliverable type in your template notes.