Skip to main content
← All posts

How to normalize audio loudness for podcasts and streaming

You finish editing a podcast episode, export it, upload it to your hosting platform, and hit publish. Then a listener messages you: “Your show is way quieter than everything else in my queue. I have to crank the volume every time your episode starts.” Or the opposite — your episode blasts out of the speakers compared to the show before it. Both problems come from the same root cause: your audio’s loudness doesn’t match the platform’s target.

Loudness normalization fixes this. It adjusts the overall perceived volume of your audio to a specific target, measured in LUFS, so it sits at the same level as everything else on the platform. This guide covers what LUFS actually means, how loudness normalization differs from peak normalization, the specific targets for major platforms, and how to hit them.

Peak normalization vs. loudness normalization

If you’ve ever used a “normalize” button in an audio editor, it probably did peak normalization: find the loudest sample in the file, then scale the entire waveform up or down so that peak hits a target (usually 0 dBFS or -1 dBFS). Peak normalization is fast and simple, and it guarantees you won’t clip. But it tells you nothing about how loud the audio actuallysounds.

A file with one loud snare hit and whispering everywhere else will peak-normalize to the same level as a heavily compressed pop track. The snare pins the peak, so the quiet content stays quiet. The pop track was already loud throughout, so it barely moves. After peak normalization, both files have the same peak level, but one sounds dramatically quieter than the other.

Loudness normalizationsolves this. Instead of looking at the single loudest sample, it measures the overall perceived loudness of the entire file — closer to how human ears actually experience volume — and adjusts to a target. This is the approach every major streaming and podcast platform uses, and it’s defined by the EBU R128 standard.

MakeMySounds has both. The volume normalizerdoes peak and RMS normalization client-side in your browser — fast and private, good enough for quick adjustments. The loudness normalizerdoes full EBU R128 LUFS normalization server-side via ffmpeg — accurate and platform-compliant.

What LUFS means

LUFS stands for Loudness Units relative to Full Scale. It’s the unit defined by the EBU R128 standard (and its ITU counterpart, ITU-R BS.1770) for measuring perceived loudness. A few key points:

  • LUFS is always negative.0 LUFS would be digital full scale — the absolute maximum. Real-world audio sits well below that. A podcast episode might measure -18 LUFS. A loud mastered pop song might hit -8 LUFS.
  • Lower numbers mean quieter. -24 LUFS is quieter than -14 LUFS. Each 1 LUFS change corresponds to roughly 1 dB of perceived loudness.
  • LUFS accounts for human hearing. The measurement applies a frequency weighting (K-weighting) that models how our ears are more sensitive to mid-range frequencies than low bass or very high treble. A bass-heavy track and a vocal-heavy track at the same RMS level will measure differently in LUFS, reflecting the fact that they actually sound different in loudness.
  • Integrated LUFS measures the whole file.EBU R128 defines several measurement windows: momentary (400 ms), short-term (3 s), and integrated (the entire program). When platforms specify a target, they mean integrated loudness — the average perceived loudness of the entire track or episode.

You might also see “LKFS” — Loudness K-weighted relative to Full Scale. For practical purposes, LUFS and LKFS are the same thing. LUFS is the EBU term, LKFS is the ITU term.

Platform targets

Every major platform that plays audio has a loudness target. If your file doesn’t match, the platform either normalizes it for you (sometimes poorly) or your content sounds off next to everything else. Delivering at the right level gives you control over how your audio sounds.

Spotify: -14 LUFS

Spotify normalizes all tracks to -14 LUFS by default (users can change this in settings, but most don’t). If your track is louder, Spotify turns it down. If it’s quieter, Spotify turns it up — but turning up a quiet track can expose noise and thin mixes. Deliver at -14 LUFS with a true peak below -1 dBTP. This applies to both music and podcast episodes hosted on Spotify.

Apple Podcasts: -16 LUFS

Apple’s requirement for podcasts is -16 LUFS with a -1 dBTP true peak ceiling. This is slightly quieter than Spotify’s target. If you distribute to both platforms, -16 LUFS is a reasonable compromise — Spotify will bump it up by 2 LUFS, which is a small adjustment that won’t cause problems.

YouTube: -14 LUFS

YouTube normalizes audio to roughly -14 LUFS. It only turns loud content down; it does not turn quiet content up. If your audio is quieter than -14 LUFS, it stays quiet. This means delivering at -14 LUFS ensures you’re as loud as YouTube allows without being turned down.

Amazon Music: -14 LUFS

Amazon Music uses the same -14 LUFS target as Spotify. Deliver at -14 LUFS with true peak below -1 dBTP.

Broadcast (EBU R128): -23 LUFS

European broadcast follows EBU R128 strictly: -23 LUFS ± 1 LU, true peak -1 dBTP. This is considerably quieter than streaming targets. If you’re producing audio for TV, radio, or film in Europe, this is the standard. US broadcast (ATSC A/85) uses -24 LKFS, which is almost the same thing in a slightly different measurement framework.

Quick reference table

  • Spotify (music & podcasts): -14 LUFS, -1 dBTP
  • Apple Podcasts: -16 LUFS, -1 dBTP
  • YouTube: -14 LUFS, -1 dBTP
  • Amazon Music: -14 LUFS, -1 dBTP
  • Tidal: -14 LUFS, -1 dBTP
  • EBU R128 (European broadcast): -23 LUFS, -1 dBTP
  • ATSC A/85 (US broadcast): -24 LKFS, -2 dBTP

The EBU R128 standard

EBU R128 is the loudness recommendation published by the European Broadcasting Union. It replaced the old practice of mixing to peak levels, which led to the “loudness war” — producers making tracks louder and louder to stand out, at the cost of dynamic range and listener fatigue.

The standard defines three things: a measurement method (ITU-R BS.1770, using K-weighted gated loudness), a target (-23 LUFS for broadcast), and a true peak limit (-1 dBTP to prevent inter-sample clipping in lossy codecs). Streaming platforms adopted the same measurement method but chose louder targets (-14 to -16 LUFS), since headphone and phone listening tolerates — and even benefits from — a louder baseline than living room speakers.

The “gated” part matters. EBU R128’s integrated loudness measurement ignores sections of audio below -70 LUFS (absolute gate) and then also ignores sections more than 10 LU below the ungated average (relative gate). This means silence and very quiet passages don’t pull down the measured loudness. A podcast with a 10-second silent intro won’t measure artificially quiet compared to one that starts immediately.

How to normalize with MakeMySounds

The loudness normalizeruses ffmpeg’s loudnorm filter, which implements the full EBU R128 algorithm: K-weighted measurement, gating, true peak limiting, and two-pass processing for accurate results.

  1. Go to the loudness normalizer page and drop your audio file. Any format works — MP3, WAV, FLAC, M4A, OGG, whatever you have.
  2. Pick a target LUFS. The page has presets for Spotify (-14), Apple Podcasts (-16), broadcast (-23), and US broadcast (-24). Or type a custom value.
  3. Click normalize. The file uploads to the server, ffmpeg measures the current integrated loudness and applies the adjustment, and the normalized file downloads automatically.

This tool uploads your file to the server for processing because accurate LUFS measurement requires the full EBU R128 algorithm, which isn’t practical to run in the browser. The file is deleted after processing — it’s not stored.

Tips for getting loudness right

  • Normalize as the last step.Do all your editing — trimming, noise reduction, EQ, compression — before normalizing loudness. Every edit changes the measured LUFS, so normalizing early just means you’ll need to do it again.
  • Don’t over-compress to hit the target. If your podcast episode measures -22 LUFS and your target is -16, the normalizer adds 6 dB of gain. If the dynamic range is huge (quiet whispers and loud laughs), the loud parts might clip. The right fix is dynamic compression (a compressor plugin in your DAW) to even out the levels beforenormalizing, not cranking the normalizer and hoping for the best. For already-mixed content that’s just a bit quiet, the normalizer handles it cleanly.
  • True peak matters.Lossy codecs (MP3, AAC, OGG) can produce peaks between samples that exceed the highest sample value. A file peaking at -0.5 dBFS might actually hit +0.3 dBTP after lossy encoding, causing distortion on playback. That’s why every platform specifies a true peak limit of -1 or -2 dBTP. The loudnorm filter respects this.
  • Pick one target and stick with it.If you distribute to multiple platforms, pick the quietest target among them. For most podcasters, that’s -16 LUFS (Apple Podcasts). Spotify will turn it up slightly, which is harmless. Going the other way — delivering at -14 and having Apple Podcasts turn it down — also works, but gives you less headroom.
  • Mono podcasts are fine.Loudness normalization works identically on mono and stereo files. Single-speaker podcasts don’t need stereo — mono at the right LUFS target sounds great and halves the file size.

When to use simple normalization instead

Full LUFS normalization is essential for content going to streaming platforms or broadcast. But sometimes you just need a quick volume boost or reduction and don’t care about LUFS compliance. For those cases, the volume normalizerruns entirely in your browser with no upload. It does peak or RMS normalization — simpler and faster, though not perception-weighted like LUFS. Good for adjusting a voice memo before sending it, evening out volume between clips you’re about to merge, or any situation where platform compliance isn’t the goal.

Ready to try it?

Open the tool →