How loud should your master be? LUFS without the myths

Your master should be loud enough to support the music and its real listening context, but not so loud that limiting damages the qualities you want to preserve. There is no compulsory LUFS number for every release. Streaming normalization targets describe playback behaviour; they are not universal creative mastering laws.

Use references from a related style, compare them at matched playback level, and meet any actual delivery specification.

What LUFS measures

LUFS is a loudness unit relative to full scale. Measurement based on ITU-R BS.1770 applies frequency weighting and channel weighting to estimate programme loudness more meaningfully than a simple peak or RMS meter.

Common views include:

• Momentary loudness for rapid changes.
• Short-term loudness for phrases and sections.
• Integrated loudness across the measured programme.
• Loudness range for the distribution of loudness over time.

Integrated LUFS cannot tell you whether the kick is punchy, the vocal is clear or the chorus feels exciting. Two songs can share a reading and behave very differently.

Why -14 LUFS is not a mastering commandment

Spotify currently uses -14 LUFS for its Normal playback setting. A track above that level is generally turned down during normalized playback. That does not mean Spotify requires every uploaded master to measure exactly -14 LUFS.

Listeners can encounter other normalization settings, devices, web players, downloads, clubs, radio, videos and normalization-off playback. Platforms can also change their behaviour. Mastering purely to one current service value confuses a playback policy with an artistic decision.

Use the platform guidance where relevant, but keep the source master musically sound on its own.

What normalization changes

Basic loudness normalization applies gain during playback:

• a louder track can be turned down;
• a quieter track can be turned up when headroom permits;
• album normalization can preserve intentional level relationships;
• track normalization can treat a song independently in a playlist.

Turning a crushed master down does not reconstruct the transient peaks that were removed. Turning a dynamic master up does not automatically make it crushed. Normalization changes level, not history.

Decide loudness in the right order

1. Finish the tonal and dynamic balance

Low-frequency excess can make the limiter work harder without creating useful perceived loudness. Harsh upper mids can become exhausting as level increases. Solve arrangement, mix balance and resonances before asking the limiter to create the final result.

2. Choose appropriate references

Use professionally released tracks with related genre, density and arrangement. Avoid treating one unusually loud release as the law for an entire style.

3. Level-match the comparison

A version only 1 dB louder can seem clearer and more exciting. Turn the louder version down before judging punch, depth, distortion and tonal balance.

4. Increase level gradually

Drive the limiter in small steps. Listen to transients, vocal consonants, cymbals, bass sustain and stereo depth. Note where damage begins. The loudest acceptable version is often slightly below the point where the meter says you can still go further.

5. Measure the whole programme

Reset integrated measurement at the start and play the complete export. Inspect section changes, loudness range, true-peak maximum and the amount of limiting. Do not infer the entire master from the loudest chorus loop.

6. Verify the exported file

Re-import the final file, measure it, and listen for conversion problems. When lossy delivery is likely, audition an encoded version or codec preview.

A useful decision framework

Ask these questions instead of “What LUFS number wins?”:

• Does the chorus still lift?
• Does the kick retain shape at matched volume?
• Is the vocal stable without becoming flat?
• Does the bass trigger audible pumping?
• Does the master remain comfortable for the song’s duration?
• Does it meet the destination’s true-peak and loudness rules?
• Does it hold up beside references after level matching?

If the answer is yes, the number is describing a successful decision—not making it.

When a strict target is appropriate

Broadcast, advertising, cinema and commissioned delivery can have explicit specifications. EBU R 128, for example, recommends -23 LUFS for programme loudness in its broadcast context and includes loudness range and maximum true-peak descriptors.

Meet a contractual specification accurately. Do not copy that broadcast target into unrelated music delivery, just as you should not copy a streaming service setting into every broadcast master.

Common loudness mistakes

• Mastering every style and arrangement to one number.
• Turning down the monitoring volume as limiting damage increases.
• Comparing processed and unprocessed audio at different loudness.
• Watching integrated LUFS while ignoring momentary behaviour.
• Using sample peak where a true-peak delivery limit applies.
• Assuming normalization makes mastering quality irrelevant.

SPC Limit Breaker combines true-peak limiting, eight modes, LUFS and true-peak metering, Auto Gain, and Unity and Delta monitoring for level-conscious decisions. For an independent end-of-chain view, SPC SpectraScope provides momentary, short-term and integrated LUFS, loudness range and true peak through an analysis-only path.

Read true peak vs sample peak before choosing a delivery ceiling, then address low-frequency efficiency with how to make bass louder without clipping.