The Handoff Document That Crashed the Delivery
A post-production sound editor finishes the dialogue edit on a half-hour television drama. They send the pre-mix to the re-recording mixer with a note: "Targeting -18 dBRMS, which is what the facility has used for broadcast for years." The mixer uses the RMS target as a guide and hits it. The mix gets delivered to the broadcaster. QC flags it as 5 LUFS over their -23 LUFS integrated requirement. The show's first air date is in six days. The re-mix runs two days and costs the facility its delivery bonus.
Nobody made an error in the mix itself. The error was in the handoff spec. The broadcaster's QC team measures integrated LUFS per EBU R128. Delivering an RMS number to a LUFS standard is not a translation problem. It is delivering the wrong metric entirely.
This post explains how RMS and LUFS are calculated differently, why they produce different readings from the same file, and exactly what to write in a handoff document so your mixer has a single unambiguous number to hit.
The loudness targets in this post are drawn from the ITU-R BS.1770-4 loudness algorithm, the EBU R128 recommendation, the ATSC A/85 broadcast standard, and current Netflix, Spotify, YouTube, and Apple Music delivery specifications as of August 2026.
How RMS and LUFS Are Calculated Differently
Both RMS and LUFS measure average audio level, but they make different assumptions about what "average" means.
RMS (Root Mean Square) measures the mathematical average power of the signal over a short time window, typically 300ms. The formula: square all sample values, take their mean, take the square root. The result is expressed in dBFS. RMS applies no frequency weighting. A 50Hz bass rumble and a 3kHz speech frequency at the same RMS level sound very different in perceived loudness because human hearing responds differently at those frequencies. RMS has no mechanism to account for that difference.
LUFS (Loudness Units Full Scale) builds on the core RMS calculation but adds three things. First, K-weighting, a frequency curve that attenuates low frequencies and slightly boosts high-mids, designed to approximate human hearing response. Second, three separate time windows: momentary (400ms), short-term (3 seconds), and integrated (full program duration). Third, gating for the integrated measurement, which excludes very quiet passages below -70 LUFS from the running average. A feature film's long silences do not pull the integrated loudness down and create an artificially quiet reading.
The practical result is that a mix heavy with bass content (LFE, action effects, kick drums) reads quieter in LUFS than in RMS, because K-weighting reduces the contribution of those frequencies. A dialogue-heavy mix with minimal bass reads more similarly in both systems. For any mix, the difference between RMS and LUFS is not a fixed conversion factor. It depends entirely on the spectral content of the material.
Three Real-World Examples
Example 1: Dialogue-heavy indie film
The mix is -14 LUFS integrated. The RMS is -18. The platform normalizes to -14, so it stays untouched. The director is happy because the dialogue level matches the platform target and the dynamic range is preserved.
Example 2: Action short with heavy bass
The RMS is -12 because of explosions. The LUFS is -16 because the bass is weighted down. The platform normalizes to -14, so the track gets turned up and the dialogue becomes too loud. The mixer has to deliver a new master that balances the bass with the platform target.
Example 3: Podcast-style interview
RMS is steady at -20. LUFS is -23 because of the gating. The broadcaster rejects it for being too quiet even though it sounds fine in the edit suite. The gap between the two numbers is the giveaway: the gating removed the quiet pauses that RMS was still averaging.
LUFS, RMS, and dBFS Compared
The table below shows what each measurement actually reads and where it fits in the workflow.
| Measurement | What It Measures | Best For | Limitation |
|---|---|---|---|
| dBFS | Highest sample value | Preventing clipping, headroom check | Says nothing about perceived loudness |
| RMS | Average energy over a window | Older workflows, rough loudness guess | Ignores how human hearing weights frequencies |
| LUFS | Perceived loudness with K-weighting and gating | Streaming, broadcast, matching levels across content | Requires a full program or long segment to measure integrated |
LUFS is the only number that matches what streaming and broadcast platforms measure during QC. RMS is useful inside the mix session for setting relative balances. dBFS is a peak warning, not a loudness value.
How to Write a Loudness Handoff Document
- Set the delivery target before mixing. Identify the platform or broadcaster and find its current spec.
- Use an LUFS meter on the master bus. Make sure it supports the EBU Mode (Momentary, Short-term, Integrated) for broadcast work.
- Monitor the integrated LUFS for the whole program, not short-term peaks. Reset the meter, play the full program, and read the integrated value at the end.
- Set a true peak ceiling of -1 to -2 dBTP to avoid intersample clipping.
- Hand the mixer the target integrated LUFS and the true peak limit in writing.
- Check the final export with the same meter used for the mix.
At the end of this process the mixer has one unambiguous number per version, specified in the same unit the QC process will use to evaluate it.
Pro Tips and Common Mistakes
Pro Tip: Always ask the platform for the exact spec: integrated LUFS, short-term max, and true peak ceiling. Three numbers, not one. A delivery sheet with only the integrated LUFS can still fail if the short-term or true peak is out of spec.
Pro Tip: Use a loudness meter that supports the EBU Mode (Momentary, Short-term, Integrated) for broadcast work. The integrated value is the only one that matters for delivery, but the momentary and short-term readings help you spot loud sections during the mix.
Pro Tip: True Peak and RMS peak are different measurements. True Peak uses inter-sample peak detection to catch clipping that occurs after digital-to-analog conversion. A mix that passes a standard peak meter at -0.5 dBFS can still fail true peak compliance at -1 dBTP.
Common Mistake: Handing a mixer an RMS target. RMS does not match streaming or broadcast delivery standards. No major platform checks RMS on delivery. Give them integrated LUFS and a true peak limit instead.
Common Mistake: Mastering to -9 LUFS to sound "loud" on YouTube. The platform turns it down to its -14 LUFS target and the dynamics are lost forever. The mix ends up flat and the limiting artifacts become audible after normalization.
Frequently Asked Questions
Is LUFS the same as loudness?
LUFS is an algorithm that approximates perceived loudness. It is the industry standard, but it is not identical to how every listener hears every room. Use it as a reliable target, not a perfect model of human hearing.
Why do two meters show different LUFS for the same file?
Meters may use different revisions of ITU-R BS.1770 or different gating settings. The difference is usually under 1 LU for compliant meters. If your meter is more than 1 LU off from the QC meter, check which standard it uses.
Which is more important: integrated or short-term LUFS?
Delivery always uses integrated. Short-term and momentary help you spot loud sections during the mix. A loud action scene may read -12 LUFS short-term, but that does not mean your program will deliver at -12 LUFS integrated.
Do I need to master to different LUFS for different platforms?
Yes, if you control the deliverables. For most indie work, -14 LUFS integrated covers YouTube and Spotify, but Netflix uses -27 LUFS, Amazon and Apple use -24 LUFS, and the BBC has its own spec. Mix each version separately.
What does K-weighting actually do?
It filters the audio to mimic human hearing. Bass is reduced and high frequencies above 2kHz get a small boost. That is why a bass-heavy mix can read high in RMS but low in LUFS.
Related Tools
The LUFS Loudness Calculator measures integrated, short-term, and momentary LUFS along with true peak for any audio file and returns a delivery-ready readout against any platform spec. For the full audio delivery workflow, Audio Delivery Standards for Film and Television covers every major platform and broadcaster. For sample rate and bit depth decisions that affect loudness measurement, Sample Rates in Film and Audio Post covers the foundational specifications.
For the authoritative technical sources, the EBU R128 loudness specification and ITU-R BS.1770-4 are the primary references.
Conclusion
RMS is useful for setting relative levels inside a mix session. LUFS is the delivery currency. Every major broadcast network, streaming video platform, and music streaming service specifies delivery in integrated LUFS, not RMS. Giving a mixer an RMS target requires a spectral-content-dependent conversion that introduces inaccuracy. Giving them an integrated LUFS target with a true peak limit gives them one number to hit against the same meter the QC process will use.
This post covers stereo and standard 5.1 surround deliveries. Dolby Atmos, Auro-3D, and spatial audio formats have additional loudness measurement considerations that differ from the ITU-R BS.1770 framework.
What LUFS target do you deliver to, and have you ever had a platform reject a mix for loudness?
