What does LUFS mean and why not just use dB?
LUFS stands for Loudness Units relative to Full Scale. It is a perceptual measurement defined in ITU-R BS.1770 that weights frequencies the way human ears do and integrates over time, unlike raw dBFS which only reports sample peaks. Two tracks at the same dBFS peak can sound very different in loudness — a hot pop song versus a sparse podcast — but LUFS gives them comparable numbers. Streaming platforms normalise uploads to LUFS targets so all content plays back at similar volume, ending the loudness wars of the 1990s and 2000s.
What LUFS target should I use for each platform?
Spotify and YouTube normalise to -14 LUFS integrated, Apple Music to -16, Amazon Music to -14, Tidal to -14, Twitch to around -16, broadcast TV (EBU R128) to -23, and US broadcast (ATSC A/85) to -24. Submit louder than the target and the platform turns you down; submit quieter and it leaves you quiet (most do not turn you up). Podcasts commonly aim for -16 LUFS with a -1 dBTP true-peak ceiling. Audiobooks for ACX require -18 to -23 LUFS with peaks under -3 dBTP. Check noise context at https://devicecheckonline.com/microphone/noise.
What is the difference between momentary, short-term, and integrated?
Momentary is a 400 ms sliding window — the most reactive reading, useful for spotting sudden loud bursts. Short-term is a 3-second window — good for monitoring while mixing because it tracks musical phrasing. Integrated is the average over the entire measurement, with gating that ignores silence and very quiet passages. The integrated value is the one that platforms use; momentary and short-term are tools for getting there. A talk show with quiet pauses might read -20 LUFS short-term during speech but -16 LUFS integrated overall.
Why does my integrated LUFS keep changing as I talk?
Because integrated LUFS includes everything you have said since the meter started. The first few seconds swing wildly because there is little data to average. After 30–60 seconds of continuous speech, the integrated value stabilises within 0.5 LU. Long pauses or silences are excluded by the absolute (-70 LUFS) and relative (-10 LU below integrated) gates defined in BS.1770-4, so a podcast with long quiet edits still reads correctly. Restart the measurement at the start of each take for an accurate integrated value.
How is LUFS different from RMS or peak?
Peak is the highest single sample, in dBFS — useful for avoiding clipping, useless for perceived loudness. RMS is the average power, which is closer to perceived loudness but does not account for hearing's frequency sensitivity. LUFS adds K-weighting (a high-pass filter plus a high-shelf boost around 2 kHz) that mimics how our ears emphasise upper midrange, then applies the BS.1770 mean-square calculation plus gating. Result: LUFS matches what listeners actually perceive as loud, across speech, music, and sound effects equally. Spot peaks live at https://devicecheckonline.com/microphone/waveform.
Why does my podcast sound quiet even at -14 LUFS?
LUFS measures the whole programme; perceived loudness in any moment also depends on dynamic range. A -14 LUFS podcast with 20 dB of dynamic range (whisper to shout) will feel quiet during the whispers, even though the average matches a -14 LUFS pop song with only 6 dB of dynamic range. Add gentle compression (3:1 ratio, -18 dB threshold) to reduce dynamic range before measuring final LUFS. Spoken word benefits from 8–10 LU loudness range; music from 6–15. Confirm room input is clean using the noise floor test first.