Loudness and LUFS Explained for Mastering

Loudness is the single most misunderstood number in mastering. Two songs can peak at the same level on a meter and still sound very different in volume, because peaks only describe the tallest moment in the file, not how the listener experiences it. LUFS, short for Loudness Units relative to Full Scale, was designed to measure perceived loudness over time, and it has become the common language between producers, broadcasters and streaming platforms.

This guide explains what LUFS actually measures, why it replaced simple peak checks as the main reference, and how to use it when you finish a track. The goal is not to chase a single number. The goal is to understand what the number means so you can make informed decisions about dynamics and level.

Why peak meters mislead

A peak meter shows the highest sample value in a short window. That information matters for avoiding clipping, but it says almost nothing about how loud a track feels. A kick drum that hits hard for a few milliseconds can push the peak meter to the top while the rest of the song sits quietly. Another track with a steady, dense mix can have a lower peak and still sound louder to most listeners.

Human hearing integrates energy over time rather than reacting to single peaks. Loudness meters are built around this idea. They weight frequencies to roughly match how the ear responds, then average the energy over a window of time. The result is a figure that tracks perceived level more closely than any peak reading.

What integrated LUFS means

The most useful number for a finished master is integrated loudness. It is the average loudness of the whole program, calculated with gating so that silent passages do not drag the figure down. A four-minute track gets one integrated value, and that value is what platforms and broadcast standards compare against their targets.

Short-term and momentary readings are useful during mixing and arrangement. Short-term uses a three-second window, and momentary uses a 400 millisecond window. These values rise and fall with the music, so they show you where the loud sections are, but you should not set a final target from a single momentary reading.

The standard that defined this approach is ITU-R BS.1770, which describes the measurement method, and EBU R128, which builds on it for broadcast delivery. EBU R128 uses a reference target of minus 23 LUFS integrated for broadcast programs, with a tolerance of plus or minus one loudness unit. Many people meet this standard through television and radio, so it appears in many production workflows.

Streaming targets and normalization

Streaming platforms use loudness normalization. When a track plays, the service measures its integrated loudness and turns it up or down so that playback lands near the platform's reference level. The exact reference differs between services, and they change their rules over time, so always check current platform documentation before finalizing a master.

Historically, many services used a reference around minus 14 LUFS, and that figure still appears often in guides. Other services have used targets closer to minus 16 LUFS. Because the platform does the adjustment, a master that is louder than the reference is not made louder to the listener. It is turned down, and any dynamic detail that was squeezed in the process stays squeezed.

This changes the logic of mastering. Pushing a track to a very high integrated level does not earn a louder result on the platform. It only costs dynamic range. A master that sits near the platform reference usually translates well, because the platform does little or no gain change.

True peak and why it matters

Integrated loudness tells you about average level, but it does not guarantee that the audio will stay clean at playback. Digital peaks can fall between samples. When the signal is converted to analog or resampled during playback, those inter-sample peaks can exceed full scale and cause distortion. This is why the true-peak measurement exists. It estimates the peak level between samples by oversampling the signal before measuring.

A common practice for streaming delivery is to keep the true-peak ceiling at or below minus 1 dBTP, which leaves a small safety margin. Many limiters and meters offer a true-peak mode, and enabling it is a simple way to catch problems that a sample-peak meter would miss.

Working with loudness in practice

Start your mastering session with a reference track that you know well. Set your meter to show integrated loudness, short-term loudness and true peak. Play the reference and note its integrated value, then play your master and compare. You are not trying to match it exactly. You are checking whether your balance of loud and quiet passages feels similar.

Make small gain adjustments rather than large jumps. A change of one or two decibels can be heard clearly, and it is easier to judge whether the change helped the song. If you find that you need heavy limiting to reach a target, step back and check the mix. Overly dense arrangements, unbalanced low frequencies or a dominant kick drum can all force the limiter to work harder than it should.

Check the master at several listening levels. Loudness perception changes with volume, so a track that sounds balanced at a moderate level may feel bright or thin when played very quietly. Checking at more than one level helps you find problems early.

Common mistakes

The most frequent mistake is treating the loudest possible result as the best result. A master pushed far beyond the platform reference will often sound fatiguing and lose the contrast between sections. Another mistake is mastering to a meter alone without listening in context. Numbers should guide your decisions, but your ears decide whether the song works.

Comparing loudness targets

The table below summarizes the common reference points you will encounter. Treat these values as starting points and confirm them against the current documentation for each platform or broadcaster.

Context	Typical integrated target	True-peak guidance
EBU R128 broadcast	minus 23 LUFS, plus or minus 1 LU	minus 1 dBTP
Streaming (common reference)	around minus 14 LUFS	minus 1 dBTP
Some services in past guides	around minus 16 LUFS	minus 1 dBTP
Club or loud genre master	varies by genre	minus 0.3 to minus 1 dBTP

The club row is intentionally vague. Loud genres may accept higher integrated levels on purpose, but the platform normalization still applies on streaming services.

Key takeaways

Loudness is perceived energy over time, and LUFS measures it. Peak meters only show the tallest moment in the file, so they cannot tell you how loud a track feels on its own.

Integrated LUFS gives one average value for the whole program. Use it to compare your master against a reference and against platform guidelines, not as a number to maximize.

Streaming platforms normalize playback. Pushing a master far above the reference does not make it louder for listeners; it only removes dynamic range. Aim near the reference and keep the true-peak ceiling at a safe margin.