---
title: "How to set LUFS and true-peak targets: matching -16 LUFS mono and -1 dBTP to avoid clipping artifacts"
site: "audobox.com"
kind: "article"
author: "Hannah Morgan"
date: "November 22, 2026"
canonical: "https://audobox.com/blog/how-to-set-lufs-and-true-peak-targets-matching-16-lufs-mono-and-1-dbtp-to-avoid-clipping-artifacts.php"
markdown: "https://audobox.com/blog/how-to-set-lufs-and-true-peak-targets-matching-16-lufs-mono-and-1-dbtp-to-avoid-clipping-artifacts.php/index.md"
description: "Mono audio passes only at -16 LUFS gated integrated, -1 dBTP true peak, and decode-back true peak, all measured at 4x oversampling or higher."
---

# How to set LUFS and true-peak targets: matching -16 LUFS mono and -1 dBTP to avoid clipping artifacts

Hannah Morgan · November 22, 2026

> Mono audio passes only at -16 LUFS gated integrated, -1 dBTP true peak, and decode-back true peak, all measured at 4x oversampling or higher.

| Takeaway | Detail |
| --- | --- |
| A mono deliverable passes only when gated integrated loudness hits -16 LUFS, true peak stays at or below -1 dBTP, and decode-back true peak also passes. | All three checks must pass on the mono file; a sample-peak-only -1 dBFS ceiling is not equivalent and can still clip. |
| Set the limiter to -1 dBTP with at least 4x oversampling before measuring. | True peak must be measured at 4x oversampling or higher; sample peak alone does not satisfy the -1 dBTP target. |
| Re-measure loudness and true peak after the delivery codec, not just on the render. | Decode-back true peak is a required pass condition alongside gated integrated loudness and oversampled true peak. |
| Re-render if any check misses by more than 0.5 LU or 0.3 dB. | Tolerance is 0.5 LU on integrated loudness and 0.3 dB on true peak; exceeding either triggers a re-render. |

This guide shows how to hit the 2026 -16 LUFS mono and -1 dBTP targets without clipping artifacts.

The answer changes once you measure true peak at 4x oversampling and re-check it after the delivery codec, since a sample-peak-only ceiling can still clip.

![How to set LUFS and true-peak](https://static.mm-ais.com/article-images-ai/how-to-set-lufs-and-true-peak-targets-ma-ai-216e9a3e.jpg)

## How LUFS and true peak actually diverge

The integrated LUFS reading on a mono file is a gated measurement, meaning it ignores silence and low-level passages to calculate an average loudness over the program’s duration. A mono deliverable can display -16 LUFS integrated while individual sample values still climb above -1 dBTP. This happens because LUFS integrates the entire waveform envelope, whereas true peak detection looks at the highest inter-sample excursions that a standard sample-peak meter might smooth over.

True peak estimation requires oversampling—typically at least 4x the original rate—to reconstruct the analog waveform between discrete samples. A sample-peak meter set to -1 dBFS can show a reading of exactly -1.0 dBFS, yet the true peak, once oversampled, may exceed that ceiling and cause clipping after decode. For this reason, the limiter ceiling must be specified in dBTP, not dBFS, because the sample-peak value is a myth that does not survive the reconstruction filter of the delivery codec.

Mono delivery concentrates all audio energy into a single channel, which changes how both the LUFS meter and the true peak meter read the material. When measuring for broadcast or streaming targets, the mono file must be processed as mono, not as dual-mono stereo, because channel weighting algorithms treat the two channels differently and platform normalization algorithms often sum channels before applying the loudness target. A mono file that passes -16 LUFS on a stereo meter may actually sit outside the loudness window when the platform sums to mono.

To meet the 2026 deliverable specification, three checks must converge: the gated integrated LUFS must read -16 LUFS (±0.5 LU), the true peak must be measured at -1 dBTP with at least 4x oversampling, and a decode-back true peak test must confirm the final file remains under that ceiling. If any of these three meters misses its target by more than 0.5 LU or 0.3 dB, the render must be re-done. A sample-peak-only -1 dBFS ceiling is not equivalent and can still clip, so the -1 dBTP limit with oversampling is the only ceiling that guarantees the file will not exceed the target after decoding.

![How LUFS and true peak actually diverge — How to set LUFS and true-peak](https://static.mm-ais.com/article-images-ai/how-to-set-lufs-and-true-peak-targets-ma-ai-99d55ce3.jpg)

## Evidence: three checks must converge

A mono deliverable meets the 2026 -16 LUFS / -1 dBTP target only when three independent measurements converge on the same file. This three-meter convergence test treats gated loudness, reconstructed peak, and codec-round-trip peak as separate gates; a pass on one meter does not clear the others, and the file is not delivery-ready until every check passes within tolerance. The test is evidence, not assumption: each meter answers a different question, and only the combined result proves the deliverable meets the target.

| Check | Measurement | Target | Record |
| --- | --- | --- | --- |
| 1 | Gated integrated loudness on the final mono file | -16 LUFS | One value to 0.1 LU |
| 2 | True peak with at least 4x oversampling | -1 dBTP | Peak value in dBTP |
| 3 | Decode-back true peak after the delivery codec | -1 dBTP | Peak value in dBTP |

Check 1 confirms the program loudness of the finished mono render. Record the integrated reading to 0.1 LU and compare it directly to -16 LUFS; a reading of -15.4 LUFS misses by 0.6 LU, which is more than the 0.5 LU re-render threshold and requires a re-render. Check 2 confirms that reconstructed sample peaks stay within the delivery ceiling. With the limiter set to -1 dBTP and at least 4x oversampling enabled, record the true-peak value in dBTP and verify it is at or below -1 dBTP on the same file used for Check 1.

Check 3 is the decode-back measurement and the myth-killer for the claim that -1 dBFS equals -1 dBTP. Encode the mono file in the delivery codec, decode it back to PCM, and measure true peak again with at least 4x oversampling. A file that reads -1.0 dBFS on a sample-peak meter can measure -0.6 dBTP after decode-back, a 0.4 dB exceedance above the -1 dBTP ceiling. That 0.4 dB gap is the myth-killing number: sample-peak compliance does not guarantee true-peak compliance, and the decoded deliverable can still clip.

When any check misses by more than 0.5 LU or 0.3 dB, re-render rather than patching the exported file. If Check 3 rises above -1 dBTP, lower the limiter ceiling, render a new mono master, and repeat all three checks in order. The convergence rule is strict because the measurements interact: a lower ceiling can reduce integrated loudness, and a loudness correction can change peak behavior. The deliverable passes only when the same final file clears -16 LUFS integrated, holds true peak at or below -1 dBTP with at least 4x oversampling, and survives the decode-back true-peak check after the delivery codec.

## Limiter ceilings compared: dBTP wins

Three limiter ceilings are in common use for mono deliverables, and they do not produce the same file. Option A is the sample-peak ceiling: set the limiter to -1 dBFS and let it catch the loudest sample. It is the fastest to render and the easiest to explain, but it fails the true-peak test because it does not control inter-sample overs. A waveform can sit at -1 dBFS at every sample and still reconstruct above that level between samples, so a sample-peak-only ceiling is not equivalent to -1 dBTP and can still clip after conversion.

Option B is the true-peak limiter set to -1 dBTP with at least 4x oversampling. This is the winner because it directly enforces the stated ceiling rather than approximating it. The oversampling lets the limiter see the reconstructed waveform between samples, so it pulls down the peaks that a sample-peak limiter would miss. It also survives decode-back checks: when the delivered file is decoded and re-measured, the true peak stays at or under the ceiling instead of drifting upward.

Option C is normalize-only: bring the file to -16 LUFS integrated and leave peaks uncontrolled. It passes the loudness gate and fails the peak gate whenever a transient pushes true peak above -1 dBTP. A single drum hit or plosive can do this while the integrated reading stays on target, because the loudness average is gated and the peak is not.

| Option | Ceiling | True-peak test | Decode-back check |
| --- | --- | --- | --- |
| A: sample peak | -1 dBFS | Fails (inter-sample overs) | Can clip |
| B: true peak, 4x+ oversampled | -1 dBTP | Passes | Passes |
| C: normalize only | None | Fails on transients | Can clip |

## Costs and numbers that change the render

The render cost for a compliant mono deliverable is straightforward but non-trivial: one full-length render to the target loudness, followed immediately by a decode-back pass to verify the true-peak ceiling. If the decode-back measurement exceeds -1 dBTP, a second full render is required, effectively doubling the time budget for that deliverable. This “one-and-done” assumption only holds when the first true-peak check lands within 0.3 dB of the target; otherwise, the clock resets for a re-render.

Do not treat a -1 dBFS sample-peak reading as equivalent to -1 dBTP. Set the limiter ceiling in dBTP, measure true peak with at least 4x oversampling, and confirm decode-back true peak after the delivery codec. If the decode-back check is outside the -1 dBTP ceiling or the stated tolerance, adjust the limiter ceiling and re-render rather than assuming a fixed extra pad will make the file safe.

Tolerance settings define the boundary between a passing deliverable and a re-render. For integrated loudness, the method limit is ±0.5 LU; for true peak, it is ±0.3 dB. These windows account for measurement variance across different meters and the inherent uncertainty of gated integration. Tightening these limits—chasing a -16.1 LUFS target or a -1.1 dBTP reading—creates re-renders without a corresponding audible improvement, as the thresholds are defined by the measurement method, not by perceptual difference.

Finally, the -16 LUFS / -1 dBTP rule breaks down when any of the three convergence points diverges by more than the stated tolerances. If the gated integrated LUFS reads -15.8 LU, the true peak reads -0.8 dBTP, or the decode-back peak reads -0.5 dBTP, the deliverable must be rechecked or overridden. This section quantifies that the -1 dBTP ceiling is the only ceiling that, when paired with a 4x oversampled measurement and a gated LUFS read, guarantees the deliverable will survive the delivery codec without unexpected clipping.

## Where the -16/-1 rule breaks

The -16 LUFS / -1 dBTP rule assumes a single mono path, but dual-mono stereo delivery can create a measurement mismatch. When a file is authored as stereo with two independent mono channels, measure integrated loudness and true peak on the actual delivered channel layout rather than assuming the mono reading transfers. If the platform sums channels or applies its own normalization, verify the final delivered file against the -16 LUFS integrated and -1 dBTP true-peak checks before submission.

Lossy codecs can change true-peak behavior during encode and decode. The decode-back true-peak check is therefore required: encode the mono file in the delivery codec, decode it back to PCM, and measure true peak with at least 4x oversampling. If the decoded peak exceeds -1 dBTP, lower the limiter ceiling by the observed overshoot and re-render, then repeat all three checks; do not patch the encoded file.

Spoken-word programming can hit the true-peak ceiling on isolated plosives or transient consonants before integrated loudness reaches the target. In that case, do not raise the true-peak ceiling; use clip gain or gentle downward compression on the offending passages, then re-measure gated integrated loudness, 4x-oversampled true peak, and decode-back true peak after the delivery codec.

These edge cases establish that the -16 LUFS / -1 dBTP rule is a necessary but not sufficient condition for delivery compliance. A mono file that reads -16 LUFS and -1.0 dBTP on the source meter can still fail if it is dual-mono, encoded lossily, or heavily dynamic spoken word. The producer must recheck the deliverable after every format transition and after any decode-back process; if any of the three checks—gated integrated LUFS, 4x oversampled true peak, and decode-back true peak—deviates by more than 0.5 LU or 0.3 dB from the target, the render must be repeated.

## Mono file to -16/-1

Start with one mono WAV at 48 kHz/24-bit and write down two settings before you touch a fader: the meter's mode (gated integrated, not momentary or short-term) and the true-peak oversampling factor (at least 4x). Those two logged values are the baseline for every check that follows. If the meter is set to sample peak or the oversampling is off, the numbers you collect are not the numbers the delivery spec asks for, and the worksheet below will send you in the wrong direction.

Checkpoint 1 is the loudness gate. Suppose the gated integrated reading comes back at -18.2 LUFS against a -16 LUFS target. The arithmetic is a +2.2 dB clip-gain move, applied to the clip, not the limiter output. Re-measure after the gain change rather than assuming the new reading lands at -16.0: raising input gain drives the limiter harder, so the limiter's own gain reduction shifts and the integrated result can land above or below the simple sum. Treat the first reading as a starting point, not a promise.

Checkpoint 2 is the peak gate. With loudness in range, suppose the true-peak meter reads -0.3 dBTP against the -1 dBTP ceiling. Lower the limiter ceiling by 0.7 dB to -1.0 dBTP and re-render. Do not trim the output fader to fake the ceiling; that changes level without changing limiter behavior, and the next render can drift back over. The ceiling belongs on the limiter, where it is enforced on every sample the limiter touches.

| Checkpoint | Illustrative reading | Action | Re-measure? |
| --- | --- | --- | --- |
| 1. Gated integrated | -18.2 LUFS | +2.2 dB clip gain | Yes — limiter behavior changes |
| 2. True peak (4x+) | -0.3 dBTP | Lower ceiling 0.7 dB to -1.0 dBTP | Yes — re-render first |

The worksheet values above are illustrative, not universal; your file will produce its own readings. What carries over is the order and the tolerance. After the delivery codec, re-measure both gates on the decoded file. If either check misses by more than 0.5 LU on loudness or 0.3 dB on true peak, re-render rather than patching the encoded file. A sample-peak-only ceiling at -1 dBFS is not a substitute for this pass: it can read clean while the reconstructed waveform still crosses the ceiling, which is why the oversampled meter and the decode-back check both stay in the loop.

## Worked Example: Run the Numbers

For an illustrative mono spoken-word example, assume a 48 kHz/24-bit mono WAV with a quiet intro and a loud closing call-to-action. Set the limiter ceiling to -1 dBTP with at least 4x oversampling, render, and measure gated integrated loudness and true peak. If the first pass reads -14.2 LUFS integrated and -0.6 dBTP, both checks miss, so correct and re-render before delivery rather than relying on a sample-peak-only reading.

Step one is the loudness correction. The gap between -14.2 and -16 is 1.8 LU, which exceeds the 0.5 LU tolerance, so the engineer lowers the gain feeding the limiter by 1.8 dB and re-renders. (Illustration: this is a worked example, not a universal offset.) The new pass reads -16.0 LUFS integrated. Step two is the peak check on the rendered file itself: the true-peak meter, still at 4x oversampling, now shows -1.0 dBTP, inside the 0.3 dB tolerance. Step three is the decode-back check: the file is encoded to the delivery codec and decoded again, then re-measured. The decoded file reads -16.1 LUFS and -1.2 dBTP. Both land within tolerance, so this pass is accepted.

The arithmetic that matters is the margin, not the headline number. A sample-peak ceiling of -1 dBFS can read compliant on a sample meter while the reconstructed waveform between samples exceeds the ceiling; that is why the oversampled true-peak reading and the decode-back reading are measured separately rather than inferred from the first pass. In this example the decode-back peak moved 0.2 dB below the render peak, which is inside the 0.3 dB allowance — but the engineer only knows that because the check was run, not assumed.

Here is the full ledger for this example:

The winner for this example is pass 2, and its break-even trigger is explicit: if any of the three checks had missed by more than 0.5 LU or 0.3 dB, the file goes back to render rather than being nudged with a trim. That single rule — re-render, do not patch — is what keeps the delivered mono file inside the 2026 target on the first client handoff.

## If/then rules for the final render

If the gated integrated LUFS measurement falls outside the -16 ±0.5 LU window, the deliverable fails the canonical rule before the limiter is even engaged. In this case, adjust clip gain or apply light compression *before* the true peak limiter, then re-measure the integrated loudness. Only once the gated LUFS sits within the -15.5 to -16.5 LU band should the limiter chain be engaged, ensuring that subsequent true peak adjustments do not push the program louder or quieter than the target.

If the true peak reading exceeds -1 dBTP when measured at 4x oversampling or higher, the limiter ceiling must be lowered immediately. A sample-peak meter operating at 1x cannot reconstruct the inter-sample peaks that a true peak meter reveals; trusting 1x sample-peak data is the most common path to accidental clipping after codec encoding. Lower the limiter ceiling until the oversampled true peak meter reads -1.0 dBTP or lower, then re-verify the integrated LUFS to confirm the level shift did not push the program out of the -16 LUFS target.

If decode-back true peak—measured after the delivery codec has completed its lossy compression cycle—exceeds -1 dBTP, the file must not be shipped in its current state. The inter-sample peaks that survive encoding often exceed the pre-encode true peak reading. Lower the ceiling by the exact overshoot amount observed in the decode-back measurement and re-render the file. This step is non-negotiable; a file that passes the pre-encode true peak check but fails the decode-back check will clip the moment it is played back on any consumer device.

These three if/then checks operationalize the -16 LUFS / -1 dBTP target into a repeatable render workflow: gated loudness, oversampled true peak, and post-encode verification. The mono deliverable passes only when all three checks converge on the same final file within the stated tolerances.

## Frequently Asked Questions

**What oversampling rate must the limiter use before I measure true peak?**

Set the limiter to -1 dBTP with at least 4x oversampling before measuring.

**Is a sample-peak-only -1 dBFS ceiling good enough to prevent clipping?**

A sample-peak-only -1 dBFS ceiling is not equivalent and can still clip.

**Do I need to re-measure after the delivery codec, or is the render enough?**

Re-measure loudness and true peak after the delivery codec, not just on the render.

**How much can a check miss before I have to re-render?**

Re-render if any check misses by more than 0.5 LU or 0.3 dB.

**What are the tolerance limits on integrated loudness and true peak?**

Tolerance is 0.5 LU on integrated loudness and 0.3 dB on true peak, and exceeding either triggers a re-render.

**What three conditions must a mono deliverable pass?**

A mono deliverable passes only when gated integrated loudness hits -16 LUFS, true peak stays at or below -1 dBTP, and decode-back true peak also passes.

## Quick answers

| What three checks must a mono deliverable pass? | Gated integrated loudness hits -16 LUFS, true peak stays at or below -1 dBTP, and decode-back true peak also passes. |
| --- | --- |
| How should the limiter be set before measuring? | Set the limiter to -1 dBTP with at least 4x oversampling before measuring. |
| Why is a sample-peak-only -1 dBFS ceiling insufficient? | A sample-peak-only -1 dBFS ceiling is not equivalent and can still clip. |
| When should loudness and true peak be re-measured? | Re-measure loudness and true peak after the delivery codec, not just on the render. |
| What tolerance triggers a re-render? | Tolerance is 0.5 LU on integrated loudness and 0.3 dB on true peak; exceeding either triggers a re-render. |

Also worth reading: **AI voiceover that sounds human in 2026: prosody-first 48 kHz/24-bit or -16 LUFS?**: [AI voiceover that sounds human](https://audobox.com/blog/ai-voiceover-that-sounds-human-in-2026-prosody-first-48-khz24-bit-or-16-lufs.php) · **Podcast vs YouTube Loudness: -16 vs -14 LUFS and 1 dBTP Limit**: [Podcast vs YouTube Loudness: -16](https://audobox.com/blog/podcast-vs-youtube-loudness-16-vs-14-lufs-and-1-dbtp-limit.php) · **How to A/B AI Mastering Plugins Against -14 LUFS / -1 dBTP**: [How to A/B AI Mastering](https://audobox.com/blog/how-to-ab-ai-mastering-plugins-against-14-lufs-1-dbtp.php)

### Related reading

- [How to A/B AI Mastering Plugins Against -14 LUFS / -1 dBTP](https://audobox.com/blog/how-to-ab-ai-mastering-plugins-against-14-lufs-1-dbtp.php)
- [Podcast vs YouTube Loudness: -16 vs -14 LUFS and 1 dBTP Limit](https://audobox.com/blog/podcast-vs-youtube-loudness-16-vs-14-lufs-and-1-dbtp-limit.php)
- [Fix Clipping and Distortion With AI Repair: 3 dB Threshold vs Re-Record](https://audobox.com/blog/fix-clipping-and-distortion-with-ai-repair-3-db-threshold-vs-re-record.php)
- [YouTube's Single Gain Coefficient: Evidence, Targets, Limits](https://audobox.com/blog/youtubes-single-gain-coefficient-evidence-targets-limits.php)
- [How to Choose an AI Audio Tool for Keyboard and Fan Noise in Remote Podcasts](https://audobox.com/blog/how-to-choose-an-ai-audio-tool-for-keyboard-and-fan-noise-in-remote-podcasts.php)
- [AI Noise Reduction for E-Learning: Verify Before You Commit](https://audobox.com/blog/ai-noise-reduction-for-e-learning-verify-before-you-commit.php)

### Latest

- [How to Choose an AI Audio Tool for Keyboard and Fan Noise in Remote Podcasts](https://audobox.com/blog/how-to-choose-an-ai-audio-tool-for-keyboard-and-fan-noise-in-remote-podcasts.php)
- [How to A/B AI Mastering Plugins Against -14 LUFS / -1 dBTP](https://audobox.com/blog/how-to-ab-ai-mastering-plugins-against-14-lufs-1-dbtp.php)
- [AI Noise Reduction for E-Learning: Verify Before You Commit](https://audobox.com/blog/ai-noise-reduction-for-e-learning-verify-before-you-commit.php)

Canonical: https://audobox.com/blog/how-to-set-lufs-and-true-peak-targets-matching-16-lufs-mono-and-1-dbtp-to-avoid-clipping-artifacts.php
Markdown: https://audobox.com/blog/how-to-set-lufs-and-true-peak-targets-matching-16-lufs-mono-and-1-dbtp-to-avoid-clipping-artifacts.php/index.md

```json
{"@context":"https://schema.org","@type":"BlogPosting","headline":"How to set LUFS and true-peak targets: matching -16 LUFS mono and -1 dBTP to avoid clipping artifacts","url":"https://audobox.com/blog/how-to-set-lufs-and-true-peak-targets-matching-16-lufs-mono-and-1-dbtp-to-avoid-clipping-artifacts.php","description":"Mono audio passes only at -16 LUFS gated integrated, -1 dBTP true peak, and decode-back true peak, all measured at 4x oversampling or higher.","datePublished":"November 22, 2026","author":{"@type":"Person","name":"Hannah Morgan"},"articleBody":"| Takeaway | Detail | | --- | --- | | A mono deliverable passes only when gated integrated loudness hits -16 LUFS, true peak stays at or below -1 dBTP, and decode-back true peak also passes. | All three checks must pass on the mono file; a sample-peak-only -1 dBFS ceiling is not equivalent and can still clip. | | Set the limiter to -1 dBTP with at least 4x oversampling before measuring. | True peak must be measured at 4x oversampling or higher; sample peak alone does not s..."}
```
