Loudness and peak level answer two different questions. Loudness, measured in LUFS, tells you how loud a track feels over time. Peak level, measured in dBFS (or dBTP for true peak), tells you how close the loudest moment comes to the ceiling of your digital system. You need both numbers, and a mix can look perfect on one while failing on the other.
I'll walk through what each meter reads, why they don't convert into each other, where true peak fits, and how to use the pair while you mix and master.
TL;DR
- LUFS is K-weighted loudness averaged over time. It is the closest meter to what your ears report.
- Peak (dBFS) is the highest single sample. It tells you about clipping risk, not about how loud the track sounds.
- True peak (dBTP) estimates the level between samples, so it can read higher than the sample peak.
- There is no fixed conversion between LUFS and dBFS, but a 1 dB gain change moves loudness by 1 LU.
- For streaming, think in pairs: an integrated loudness target and a true-peak ceiling of −1.0 dBTP, or lower if you want a safety margin.
Table of Contents
- The short version: a ceiling and an average
- What peak level (dBFS) actually measures
- What loudness (LUFS) actually measures
- Why you can't convert LUFS to dBFS
- Sample peak vs true peak
- Side by side
- Why streaming cares about both
- Is −10 or −14 LUFS too loud?
- How to use both meters while you mix and master
- Our perspective
- Get both numbers, and what to do about them
- FAQ
- Sources
The short version: a ceiling and an average
The easiest way to hold the difference in your head is this. Peak is a ceiling. Loudness is an average.
Imagine two songs. One is a sparse ballad with a snare that spikes hard but sits quietly the rest of the time. The other is a dense, heavily limited pop track. Both can hit the same peak of −1 dBFS, yet the second will sound far louder, because its energy sits near the top almost all the time. The peak meter sees two identical songs. The loudness meter sees two very different ones.
That is why peak alone can't tell you how loud a track is, and why loudness alone can't tell you whether it will distort.
What peak level (dBFS) actually measures
In a digital system, 0 dBFS is the maximum possible sample level. Everything you record or bounce sits at or below it, which is why digital levels are usually negative numbers. dBFS means decibels relative to full scale.
A sample-peak meter simply reports the highest individual sample it has seen. It is instantaneous: one sample, one moment. That makes it excellent for answering "did I clip?" and nearly useless for answering "how loud is this?" A single stray transient can set the reading while the rest of the song sits 15 dB lower.
If you want to go deeper on how much room to leave while mixing, we cover it in Mix Headroom: The −3 to −6 dBFS Rule Has a Catch.
What loudness (LUFS) actually measures
LUFS stands for Loudness Units relative to Full Scale. Instead of looking at one sample, it applies K-weighting to account for how tonal balance affects perceived loudness, then averages the result over time. A bass-heavy mix and a bright mix with the same raw energy won't read the same, because the ear doesn't hear them the same.
You will see LUFS in three common flavours:
- Momentary (M): a very short window, good for watching a chorus hit.
- Short-term (S): a few seconds, handy for balancing sections while you mix.
- Integrated (I): the whole song from start to finish. This is the number streaming services care about.
One small point of vocabulary: a change in loudness is expressed in LU, and a 1 dB gain increase raises the loudness level by 1 LU. Our LUFS explained guide goes through the window types in more detail.
Why you can't convert LUFS to dBFS
People search "LUFS vs dBFS" hoping for a conversion table. There isn't one. LUFS and dBFS are different scales: one measures loudness, the other measures amplitude against digital full scale. LUFS is frequency-weighted and time-averaged, whereas dBFS is instantaneous, so the same song can have wildly different gaps between its two readings depending on how it was mixed and limited.
What you can rely on is the relative behaviour. Turn the whole track up by 1 dB and both numbers move by 1. That is what makes gain staging with both meters predictable.
The gap has a name: PLR
The distance between the two is useful in its own right. PLR (peak to loudness ratio) describes the relationship between a program's maximum true peak and its integrated LUFS. A small PLR means the track is heavily compressed and limited. A larger PLR means more dynamic range between the peaks and the average. Leapwing notes that heavily compressed music can sound constrained, which is a good reason to treat PLR as a check on your limiter, not a number to chase.
If you also want the crest factor and LRA in plain language, see our audio metrics glossary.
Sample peak vs true peak
Here's where many mixes get caught out. A sample-peak meter reads the discrete digital samples. But your listener's device converts those samples back to an analogue waveform, and that waveform can curve between samples. According to the BBC research white paper on true-peak metering, true peak estimates the analogue signal level between samples, and the reconstructed waveform can rise above the highest individual sample.
In practice, true-peak meters do this by upsampling the signal, often by a factor of four, to approximate the analogue output. The result is measured in dBTP, referenced to digital full scale. A track can show −0.1 dBFS on a sample-peak meter and still read above 0 dBTP, which means true-peak overshoots can potentially cause clipping once decoded.
This matters most on loud, limited masters, where lots of samples sit right under the ceiling. Our pieces on true peak for mastering and stopping intersample peaks with −1 dBTP headroom show what that looks like in a session.
Side by side
| Sample peak (dBFS) | True peak (dBTP) | Integrated loudness (LUFS) | |
|---|---|---|---|
| What it reads | Highest single sample | Estimated highest level between samples | K-weighted average over the whole track |
| Time scale | Instantaneous | Instantaneous | Averaged |
| Tells you about | Clipping in the file | Clipping after decoding or conversion | How loud it sounds |
| Typical use | Tracking, mixing headroom | Final master ceiling | Streaming targets, level matching |
| Can it exceed 0? | Not in the file | Yes, if inter-sample peaks overshoot | Not applicable, it's a loudness figure |
Why streaming cares about both
Streaming services use loudness normalisation so listeners don't have to reach for the volume knob between songs. On most services, a track will be turned down to about −14 LUFS, though it won't always be turned up to that level. That is why shouting louder than the target rarely pays off: the platform simply brings you back down.
Peaks come into it as well. If a song's true peak is too high, it risks being turned down by the normalisation algorithm or distorting. The AES streaming guidance recommends consistent loudness and a maximum peak of −1.0 dBTP so lossy encoders don't clip.
A word of caution about the meters themselves. One engineer writing in TV Technology points out that true-peak meters can be off by 0.5 to 1 dB because oversampling and filters affect their readings, and suggests considering limits of −2 or −3 dBTP for extra headroom. If your master is going through a lossy codec, that margin is cheap insurance.
We've written up the whole routine in −14 LUFS, −1 dBTP: how producers meter for streaming, and our streaming loudness targets article looks at how louder genres approach it.
Is −10 or −14 LUFS too loud?
Neither is "too loud" in itself. It depends on the genre, the mix and what you want it to feel like. Apple's Logic Pro Mastering Assistant documentation says −14 LUFS-I is a common target for streaming platforms, but adds that sticking rigidly to that value isn't always best and that you should master to complement the mix.
A track mastered to −10 LUFS integrated will usually be turned down on a normalising service. Whether it sounds better after that depends on what you gave up to get there: transient punch, low-end clarity, or dynamic movement. If you want to see where finished records in your genre actually land, our measured genre benchmarks and the pop mastering benchmarks are a good reality check.
How to use both meters while you mix and master
Here's the workflow I'd suggest, and it follows what the research on loudness metering recommends.
- While mixing: watch short-term LUFS. A starting point is around −20 to −18 LUFS short-term, which leaves headroom for mastering. Keep an eye on LRA so the song still breathes.
- Balance with loudness, not just faders: use the LUFS meter to compare key elements, such as vocal against the backing, rather than trusting peak meters, which jump on every transient.
- Before mastering: check the sample peak so nothing is clipping on the way in.
- While mastering: monitor integrated loudness and true peak together. Integrated loudness and True Peak are the two numbers to watch at this stage.
- Set the ceiling: put your limiter's true-peak ceiling at −1 dBTP, or lower if the track is dense or going through lossy encoding.
- Check PLR: if the gap between true peak and integrated LUFS has shrunk to almost nothing, ease off the limiter.
- Listen last: meters are guard rails. Your ears and the song decide.
Before you send anything out, a full post-production checklist helps make sure delivery, versions and sign-off don't get missed after the loudness work is done.
Common mistakes
- Reading a peak meter and assuming the track is loud enough.
- Mastering to hit −14 LUFS on the dot, then wondering why it sounds flat.
- Leaving the limiter at −0.1 dBFS sample peak and ignoring inter-sample overshoot.
- Using a mix that is too quiet and "fixing" it with a limiter. Measure first, then fix, as in our guide to a mix that's too quiet.
Our perspective
We see the same pattern again and again: producers who are anxious about one number and blind to the other. Someone chases a loudness figure and ends up with a flattened, clipped master. Someone else keeps peaks safely low and wonders why their track feels timid next to a reference.
The truth is that neither number is a quality score. They are measurements, and measurements only help when you know what they describe. Loudness tells you about perceived level, peak tells you about the ceiling, and the space between them tells you a lot about dynamics. Genre decides what a healthy gap looks like, which is why we built our evaluation around current genre standards and give every report both subjective guidance and objective measurement.
The measurements are the easy part. What matters is what you do with them: whether the low end is masking the vocal, whether the stereo image holds up, whether the limiter is working too hard. That is where experience counts, and it is why AubioMix pairs audio analysis with the practical know-how of a multi-award winning mix engineer with 20+ years in the studio.
— AubioMix
Get both numbers, and what to do about them
If you'd rather not piece this together from three plugins, upload your mix to AubioMix. You get loudness, peak and true-peak readings alongside written and visual feedback with actionable steps across 15+ areas, from compression and EQ to stereo width, phase and mastering prep. A single Mix Review is £4.99, a Reference Pass is £2.99, and the Professional plan is £19.99 per month if you review mixes regularly. You can see all options on the pricing page.
FAQ
Is LUFS the same as dBFS?
No. LUFS measures loudness, using frequency weighting and averaging over time. dBFS measures amplitude relative to digital full scale, usually at a single instant. They share a decibel-style scale, but there is no fixed conversion between them.
Is LUFS or true peak more important?
They do different jobs, so you need both. Integrated LUFS tells you how loud the track is against streaming targets, while true peak tells you whether it will clip once decoded. For a release master, set a loudness target first, then make sure the true-peak ceiling is safe.
How many dB is 1 LUFS?
A LUFS reading isn't a dB amount you can convert, but changes map neatly: increasing gain by 1 dB raises loudness by 1 LU. That is why moving your whole master up or down by 1 dB shifts both its peak and its LUFS reading equally.
Is −14 LUFS too loud?
Not usually. It is a common streaming reference, and Apple's Logic Pro documentation describes it that way while noting you shouldn't follow it blindly. Mastering a little louder or quieter can be right for the genre, as long as the track still sounds good after normalisation.
Sources
- BBC Research & Development White Paper WHP202
- What are LUFS? The complete guide (iZotope)
- What Is a True Peak Limiter? (iZotope)
- AES TD1004: Audio Streaming and Network File
- Advantages of True Peak Metering (TV Technology)
- Loudness and Dynamics (Leapwing Audio)
- What Is Loudness? (Yamaha Hub)
- Mastering Assistant parameters in Logic Pro (Apple Support)
- Your Post Production Checklist From Ingest to Archive