# What Is LUFS? The Loudness Number Streamers Ignore URL: https://synth.stream/journal/what-is-lufs Type: blog Locale: en Published: 2026-10-15 Updated: 2026-10-15 --- > LUFS measures how loud audio feels, not how high it peaks. Here is what the number means, what platforms do with it, and how to use it on stream. LUFS is a loudness unit: Loudness Units relative to Full Scale. It tells you how loud audio feels over time, not how high the loudest spike goes. If you stream, produce or upload music, it is the number that decides whether your track gets turned down, boosted or left alone. Learn it once, stop guessing at the fader. ## What is LUFS, and why do peaks lie about loudness? Peak meters show the tallest spike in the waveform. Your ears do not work like that. A kick drum that hits for 20 milliseconds can peak at -1 dBFS while the track still sounds quiet. A wall of compressed supersaws can peak at the same level and feel twice as loud. LUFS fixes that. The measurement, defined in ITU-R BS.1770, runs your audio through a filter that mimics how we hear (the K-weighting curve), then averages the energy over time. The unit is relative to digital full scale, so 0 is the ceiling and everything real is negative. One LU (loudness unit) equals one dB. Turn a track down 3 dB and its LUFS reading drops by 3. LKFS means the same thing. The EBU uses LUFS, broadcast standards in the US say LKFS. Same meter, two names. ![Streamer desk with two monitors showing waveforms and a boom mic](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/synthstream/2026-10/ede969-i1.webp) Two tracks can share the exact same peak and sit 8 LU apart. That is the whole reason LUFS exists. Broadcasters spent years dealing with ads that were far louder than the show around them, and the fix was to stop normalising to peaks and start normalising to perceived loudness. The [EBU R 128 recommendation](https://tech.ebu.ch/docs/r/r128.pdf) says it plainly: peak normalisation led to considerable loudness differences between programmes, and those differences caused most viewer complaints. Their answer was a target of -23 LUFS for broadcast and a true peak ceiling of -1 dBTP. Peaks still matter. They tell you about clipping and headroom. Loudness tells you about experience. You need both numbers, and you should stop treating them as the same thing. ## Momentary, short-term, integrated: which LUFS number do you read? Every loudness meter shows three readings. They are the same measurement over different windows. - **Momentary (M):** the last 400 ms. Jumps around. Good for watching a drop hit. - **Short-term (S):** the last 3 seconds. Good for judging a chorus or a build. - **Integrated (I):** the whole track, start to finish. This is the one platforms care about. Integrated loudness uses gating. Silence and very quiet passages are ignored, so a long ambient intro does not drag your number down. That is why a track with a 90-second breakdown can still read the same integrated value as a flat-out banger. You only get a trustworthy integrated number after playing the full track through the meter. A 20-second loop gives you a loop reading, not a track reading. Bounce first, measure second. Loudness range (LRA) is the fourth number you will see. It describes how much the loudness moves between the quietest and loudest parts. Dance music sits low on that scale. A film score sits high. It is a description, not a target. ## What the platforms actually do with your LUFS This is the part that changes how you master. Streaming services measure your track and adjust playback so everything lands near the same level. Spotify is the one that documents it clearly. Its [loudness normalization page](https://support.spotify.com/us/artists/article/loudness-normalization/) states that tracks are adjusted to -14 dB LUFS following ITU 1770. Three details from that page worth keeping in your head: - Loud masters get turned down. No extra distortion, just a lower volume, so your crushed -7 LUFS master plays at the same level as a -14 one, with less punch. - Quiet masters can get boosted, but Spotify leaves 1 dB of headroom for lossy encoding. A track at -20 LUFS with a true peak of -5 dBFS only gets lifted to -16. - Premium listeners can pick Loud (-11), Normal (-14) or Quiet (-19). Your track will meet all three. Spotify's own mastering advice: aim for -14 integrated LUFS, keep true peak under -1 dB, and if you go louder than -14, keep true peak under -2 dB to avoid encoding distortion. Different services use different targets and different rules, so do not assume one number fits everywhere. For anything you care about, read the current docs of the platform you are uploading to. ![Close-up of mixer faders and knobs under side light](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/synthstream/2026-10/78e6e8-i2.webp) ## Should you master to -14 LUFS? Skip the religion No. -14 is a reference, not a law. Plenty of club-oriented electronic music is mastered hotter, and that is a creative call. Techno that sits at -9 or -10 integrated will get turned down on Spotify, yes. It will also keep the density and aggression you built, which is why people still do it. Where it goes wrong is when you chase loudness with a limiter until the kick loses its transient. If the platform is going to turn you down 4 dB anyway, that extra limiting bought you nothing except a flatter track. My rule of thumb, and it is only that: - Streaming-first release, dynamic material: aim near -14, let the platform do nothing. - Club-first release: master for the room, check that the true peak stays under -1 dBTP, accept the turn-down online. - Anything with a quiet intro and a big drop: trust the integrated reading, not what the first 30 seconds look like. Worth the effort: A/B your master against a reference track after matching their loudness. If you compare at different volumes, the louder one always wins. That is the trap LUFS lets you escape. ## LUFS for streamers: where it bites on Twitch and Kick Here is the live angle most explainers skip. When you stream, nobody normalises your audio for you the way Spotify does. Your OBS mix goes out as you built it. I am not aware of a published LUFS target from Twitch or Kick, so treat this as a mixing problem you solve yourself. What goes wrong in practice: - Your voice sits at one loudness, your game at another, your music at a third. - A track that was mastered hot, played at the same fader position as one mastered quiet, buries your voice. - You change tracks at 2 a.m., the new one is 6 LU louder, and chat tells you about it. The fix is boring and it works. Pick a music source, play a few of your usual tracks through a loudness meter, and note where each one lands. Set your music bus so the average sits a clear step under your voice, then leave it. A fixed gap between voice and music matters more than the exact number. A limiter on the music bus catches the surprises. A compressor on the voice keeps the gap steady. Neither needs to be fancy. The OBS built-in filters are enough to start. This is also where AI music enters. Generated tracks do not come with a guaranteed loudness. One generation can land well under another from the same tool, and tracks mix at different densities. Do not assume anything about a tool's output level. Export, measure, then set your bus. Mubert is built around streaming and looping, which makes it easy to test: play an hour of it through a meter and see how far it drifts. ## How to measure LUFS without buying anything You do not need a $200 plugin. Free meters exist and they all implement the same BS.1770 standard. - Use the loudness meter built into your DAW, or a free plugin. Check that it shows integrated LUFS, short-term LUFS and true peak. - Play the whole track or the whole set from the start. Reset the integrated reading before you press play. - Read integrated, then read true peak. Write both down. - Adjust the gain at the end of your chain, not in the middle, so you are not changing how your compressors work. - Re-measure. Do not trust the gain knob. If you want the same check on a finished file, drag it into a meter that supports offline analysis. Offline analysis runs faster than real time and gives you the same integrated value. One more thing: test the file after encoding. MP3 or AAC conversion can push the true peak up by a fraction of a dB. That is why the platforms ask for headroom. ![Empty club stage with speaker stacks during soundcheck](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/synthstream/2026-10/b76196-i3.webp) ## AI-generated tracks and LUFS: check before you trust Generators like Suno and Udio hand you a finished stereo file. That file has a loudness, and nobody promised it matches the last one you made. If you are building a stream playlist or a DJ set from generated tracks, measure each one and write the integrated value in a spreadsheet. It takes a minute per track and saves you the 2 a.m. volume jump. Two cases where it matters. First, you plan to release the track: then you master it like any other, and the generator's output is a starting point. Second, you play it live: then consistency beats perfection, and a gain trim per track is enough. If you only need background music with no release plans, a royalty-free generator with steady output makes the leveling easier. Soundraw is one option built for that kind of use. Licensing is a separate question and LUFS will not answer it. Check the terms for stream use before you rely on any of these tools. ## The numbers worth writing on a sticky note - **-23 LUFS:** EBU R 128 broadcast target. - **-14 LUFS:** Spotify's normalization level, and a sane streaming reference. - **-11 and -19 LUFS:** Spotify's Loud and Quiet listener settings. - **-1 dBTP:** the true peak ceiling to respect, and -2 dBTP if you master louder than -14. - **1 LU = 1 dB:** gain changes move the reading one for one. Measure integrated, check true peak, compare against references at matched volume. If you stream, set a fixed gap between voice and music and stop touching it. That is the whole job. ## FAQ ### What does LUFS stand for? Loudness Units relative to Full Scale. It measures perceived loudness using the ITU-R BS.1770 algorithm, and one LU equals one dB. LKFS is the same measurement under a different name. ### What is a good LUFS level for music? For streaming, -14 integrated LUFS is a sane reference because Spotify normalizes to it. Club-focused masters often sit louder, and the platform simply turns them down. It is a creative choice, not a rule. ### Is LUFS the same as dB? Not quite. dBFS measures signal level, usually peaks. LUFS measures perceived loudness over time with a weighting filter that mimics human hearing. Changing gain by 1 dB changes the LUFS reading by 1 LU. ### What is the difference between integrated, short-term and momentary LUFS? Momentary uses a 400 ms window, short-term uses 3 seconds, and integrated covers the whole track with gating. Platforms care about integrated, so measure the full file. ### What LUFS should I stream at on Twitch or Kick? There is no published target to follow, so set your own. Measure your usual tracks, keep music a consistent step under your voice, and use a limiter on the music bus to catch loud surprises. ### Does Spotify lower loud songs? Yes. Spotify adjusts tracks to -14 dB LUFS during playback. Louder masters get turned down without extra distortion, and quieter ones can be lifted while keeping 1 dB of headroom.