LUFS and Loudness: Getting Perceived Volume Right

What is LUFS, and how is it different from dB?

LUFS stands for "Loudness Units relative to Full Scale." The classic peak (dBFS) meter in your DAW shows the instantaneous digital peak of the signal, but it correlates poorly with how loud your ears actually perceive it. That's because your ear isn't equally sensitive to every frequency: we hear the mids (vocals, snare, the melody region) as louder than the deep lows and the far highs. LUFS uses a filter that models this human response (K-weighting) plus time averaging to answer the real question: how loud do we actually perceive this?

LUFS is always a negative number: 0 LUFS is the theoretical ceiling, and music always sits below it (around -8 or -14). The closer you get to zero, the louder and usually the more squashed the track is. A handy rule of thumb: 1 LU is roughly equal to 1 dB, so the 5 LU gap between -14 and -9 is a clearly audible jump in loudness. The term LKFS you'll see in broadcast is the exact same thing as LUFS.

• dBFS peak = your technical ceiling (prevents clipping); LUFS = perceived loudness. Don't confuse the two.
• 1 LU is roughly 1 dB; make mastering decisions against LUFS, not the peak meter.
• LUFS is always negative; the closer to zero, the louder but the more squashed the sound.

Integrated, short-term, and momentary: which measures what?

Your LUFS meter shows three different readings, and it's easy to mix them up. Momentary measures the last 400 milliseconds; it catches instantaneous, fast peaks. Short-term measures the last 3 seconds; it tells you how loud a chorus or a section is. Integrated is the average of the entire track from start to finish (with a "gate" that ignores the silent parts). Streaming services base their decisions on this integrated value; when you say "my track is -14 LUFS," this is what you mean.

In practice you use all three together: you lock the integrated to your target (-14), you watch the energy difference between the intro and the chorus with short-term, and you use momentary to see how hard the limiter is working on the peak moments. If you only watch integrated and ignore short-term, your chorus will either vanish on a phone or blow up uncomfortably loud.

• Your streaming target = Integrated LUFS. Set it around -14.
• Use short-term to watch the balance between sections; large gaps create a "lost" feeling on small speakers.
• If momentary is jumping a lot, there's percussive transient energy; set your limiter accordingly.

Why true peak matters: the -1 dBTP rule

A normal peak meter only reads the digital sample points. But when audio leaves your phone it gets converted from digital to analog, and in that conversion the signal can rise above the actual sample value in between the sample points. This is called an inter-sample peak. A True Peak (dBTP) meter catches these hidden peaks by recalculating the signal at higher resolution (oversampling).

Why should you care? Because when the track is converted to a lossy format like MP3 or AAC, those inter-sample peaks push past 0 dBFS and create audible distortion. That's why you set your limiter's output ceiling to -1 dBTP, not 0. On bass- and 808-heavy tracks this overshoot is more aggressive, so -1.5 dBTP is safer there. The streaming service will adjust your level anyway; protecting the ceiling is on you.

• Set your limiter output to -1 dBTP and turn on True Peak mode.
• Lossy codec + 0 dBFS ceiling = audible distortion; -1 dBTP prevents it.
• On 808/sub-heavy tracks, -1.5 dBTP is safer.

Streaming normalization: why is everyone pulled to -14 LUFS?

Streaming services don't want the listener reaching for the volume every time the song changes. So they line up every incoming track to the same perceived loudness. The rough targets look like this: Spotify -14 LUFS, YouTube -14 LUFS, Amazon and Tidal around -14, and Apple Music at -16 LUFS (Sound Check). These values get updated from time to time, but the logic stays the same.

If your master is louder than that target — say -8 LUFS — the service turns it DOWN to -14, meaning it drops the gain. The extra loudness goes in the trash; but the distortion and pumping from your over-compression stay. If your master is below the target (say -20 LUFS), most services with normalization on will turn it UP toward -14. In short, the system pulls you to the average: there's no reward for going super loud, only loss.

• When the service turns you down, the distortion from extra compression STAYS while the loudness is GONE. A losing trade.
• Apple Music targets around -16; a -14 master plays fine there too, so a separate master isn't required.
• Normalization is on by default for most listeners; master for that majority.

The loudness war and why you should target -14

Back in the CD/radio era of the '90s and 2000s there was a race: "louder = more attention-grabbing." Masters got pushed toward -6 and -5 LUFS, peaks were crushed, dynamics died, and the result was tiring, airless sound. This is called the "loudness war." Streaming normalization has largely made that race pointless: everyone gets pulled to the same level anyway.

Target -14 LUFS today and you get two wins. One: the service leaves your master alone, so what you made is what plays. Two: you keep your dynamics, meaning your chorus genuinely "hits" because there are quieter moments around it to compare against. An over-crushed -8 master, once it's turned down to -14, sounds both quieter and more squashed; the worst of both worlds. Remember: loudness perception is relative, and a track with dynamics left in can feel louder than a crushed one.

• Integrated -14 LUFS = a safe, service-friendly starting point.
• Leaving dynamics in makes the chorus "hit" harder; crushing everything does the opposite.
• Genre matters: aggressive trap/metal can run a bit hotter (-10/-12), but the ceiling is still -1 dBTP.

Reaching loudness in the right order

Forcing loudness at the very end with a limiter is the most common mistake. An ordered, clean chain gives you a far better result. Start with mix balance: cut a few dB of mud between 200-400 Hz, get the kick/808 and vocal sitting clearly, and keep the low end under control. If the mix isn't right, no limiter will save it.

Then the master chain runs in this order: (1) A light bus "glue" compression — a VCA-style bus compressor, 2:1 or 4:1, slow attack (10-30 ms), medium release, only 1-3 dB of gain reduction. The goal isn't loudness, it's fusing the elements together. (2) A gentle touch-up with master EQ. (3) A True Peak limiter last: output at -1 dBTP, and bring the integrated to -14 while metering. If your limiter needs to "eat" more than 3-6 dB, stop and go back to the mix/compression; you're solving loudness in the wrong place.

Judge with your ears throughout. Don't kill the music just to hit a number; the meter is a guide, not the boss.

• If you need more than 3-6 dB of gain reduction going into the limiter, go back to the mix.
• Bus compressor = glue (1-3 dB), limiter = ceiling and final loudness. Don't conflate them.
• Match your reference to the same LUFS and A/B; otherwise you fall into the "louder = better" trap.

Club and DJ playback vs streaming

The club and DJ environment works completely differently from streaming. There's no normalization there, the sound system is already enormous, and the sub bass is physically felt. For this setting, a warmer and tighter master is usually preferred; the -8 to -10 LUFS range is common, sometimes even tighter. But true peak still has to stay at -1 dBTP, otherwise distortion sounds brutal on a big system.

The most professional approach is to bounce two separate masters from the same mix: a dynamic, -14 LUFS version for streaming, and a tighter version with controlled sub bass for club/DJ distribution. On the club master, making everything below roughly 100-120 Hz mono prevents phase issues and preserves power. A single master won't sit perfectly in every environment.

• Club/DJ master: it can be -8/-10 LUFS, but true peak is still -1 dBTP.
• Make the sub bass (below ~100 Hz) mono; it's critical for phase and power on a big system.
• Separate exports for streaming and club give a more professional result.

The phone and small-speaker reality

The vast majority of your listeners won't hear your track on studio monitors; they'll hear it on a phone speaker, a laptop speaker, cheap earbuds, or a small Bluetooth speaker. What these systems have in common: there's almost no sub bass. So the sense of loudness and clarity comes largely from the mids; the vocal, snare, melody, and the main rhythmic energy.

This is why you shouldn't judge your master only on nice monitors. Listen on a phone speaker, on a single earbud, in the car. On top of that, an over-crushed master tires the ear faster on a phone, because a small speaker already pushes the squashed midrange forward. When you A/B, always bring the reference to the same loudness; your brain automatically assumes the louder one is "better," and that illusion traps you most on small speakers.

• Listen to your master on a phone speaker, a Bluetooth speaker, and headphones separately.
• There's no bass on a small speaker; vocal and midrange clarity carry the loudness impression.
• When you A/B, match the reference to the same level (loudness-matched), or your judgment breaks down.

Summary, common mistakes, and a practical way out

Let's wrap up: LUFS measures perceived loudness, target the integrated value around -14 for streaming, keep the ceiling at -1 dBTP, bounce a separate and tighter master for the club, and listen to your master everywhere. The common mistakes are just as clear: only watching the peak meter and never measuring LUFS, pushing well above -14 and getting turned down by the service, forcing loudness with a limiter when the mix is broken, deciding only on monitors, and doing A/B without a reference (not loudness-matched).

Setting up all this measuring, the ordered chain, true peak protection, and genre-appropriate targeting by hand takes time and a calibrated ear. Getting even one part wrong can leave the result sounding thin or distorted on a phone. If you'd rather have these steps handled automatically instead of wrestling with each one, you just upload your track and pick the target platform; you can quickly get a balanced, headroom-preserving, platform-ready master. Verify what you learn with your ears, and let the repetitive work run itself.

• Pre-release checklist: Integrated ~-14, True Peak ≤ -1 dBTP, test on a phone, loudness-matched A/B against a reference.
• Don't trust the "louder" reflex; always measure.
• Bounce separate exports for streaming and club.