Measure a podcast episode, voiceover or ad read to the EBU R128 standard, then land it on -16, -14 or -23 LUFS with a true peak limiter holding -1 dBTP. You see the loudness before and after, and if the peaks are too spiky for one gain change to reach the target, the page tells you where it actually landed and why. Your file is read from your disk and never uploaded.
LUFS stands for Loudness Units relative to Full Scale. It is not peak level and it is not RMS. It is a measurement defined by ITU-R BS.1770-4 that runs your audio through a filter shaped roughly like human hearing, chops it into 400 millisecond blocks, throws away the silent ones, and averages what is left. That is why it tracks how loud something actually feels rather than how tall the waveform looks.
Every large platform now measures uploads this way and moves them toward its own figure. What they do in each direction is not the same. A file that is much louder than the target gets pulled down everywhere, so a carefully squashed master ends up flat and lifeless next to a show that was delivered correctly. A file that is quieter than the target is a coin toss: some services raise it, and raise the noise floor with it, while YouTube only ever turns loud uploads down and leaves a quiet one quiet, so it simply plays softer than everything around it. Delivering on target yourself is how you stop either from happening to you.
-16 LUFS is the podcast figure published by Apple Podcasts for stereo, with -19 LUFS for mono. -14 LUFS is the delivery figure Spotify publish, and it is roughly where YouTube, Instagram, TikTok and Facebook sit. -23 LUFS is EBU R128, the European broadcast delivery standard. Apps like Overcast and Pocket Casts publish no target at all and apply their own processing, so there is nothing to deliver to there. These are the figures the platforms publish today and they do change them from time to time, so check the spec sheet if you are delivering under contract.
Mono is the part people get wrong. A mono file played back through two speakers measures about 3 LU louder than the same signal sitting in a single channel, because the loudness standard adds the channel powers together. That is exactly why Apple Podcasts publish -16 LUFS for stereo and -19 LUFS for mono. Apple are the only ones who publish a mono figure, so this page detects the channel count, offers the 3 LU offset on the podcast preset and leaves it off everywhere else, and it never changes the channel count between measuring and writing.
Raising loudness raises peaks. A true peak, dBTP, is the level the waveform actually reaches between your samples once a converter or a lossy codec reconstructs it, which can be well above the tallest sample in the file. This page measures it by oversampling four times through a windowed sinc filter, the same method the reference EBU meters use, and the limiter holds the ceiling on that oversampled signal rather than on the raw samples. That is the difference between a file that reads clean in your editor and distorts on someone's phone, and one that does not.
Drop a file and read the numbers without exporting anything. Integrated loudness, true peak and loudness range are a complete quality check on an episode before you hand it to a client, and it takes a few seconds. Loudness range, LRA, tells you how much the level moves across the programme: a talking head interview usually lands in single digits, and a very high figure normally means one section is much quieter than the rest and needs attention that no single gain change can fix.
It is gain staging, not mastering. It applies one gain to the whole file and limits the peaks. It does not do equalisation, compression, de-essing, noise reduction or levelling between two speakers who recorded at different volumes. If one voice is buried under another, fix that first in an editor, because a single loudness number cannot repair it.