Podcast Loudness Standards: LUFS Explained (-16 LUFS and Why It Matters)

The Short Version

If you just want the numbers:

The safest single answer for a podcast distributed everywhere: -16 LUFS integrated for stereo, -19 LUFS for mono, true peaks no higher than -1 dBTP. The rest of this guide explains what those units mean, why the targets differ, and how to hit them reliably.

What LUFS Actually Is (Without the Math)

LUFS stands for Loudness Units relative to Full Scale. It's a measurement of how loud audio sounds to a human, averaged over time. One loudness unit (LU) corresponds to roughly 1 dB of level change, so the scale feels familiar — but what it measures is different from the meters you may know:

• Peak/dBFS measures the single highest sample. It tells you whether you'll clip, and nothing about how loud the audio feels. A file can peak at -1 dB and still sound whisper-quiet.
• RMS measures average electrical energy. Closer to perceived loudness, but it treats all frequencies equally — and human ears don't. A 60 Hz rumble and a 3 kHz voice at the same RMS sound wildly different in loudness.
• LUFS is RMS with two additions: a frequency weighting that mimics the ear (mid and high frequencies count more, deep bass counts less) and a gate that ignores silence and very quiet passages, so pauses between sentences don't drag your average down.

When people say "podcast loudness," they almost always mean integrated LUFS — the loudness of the entire episode averaged from start to finish. That's the number platforms measure and the number every target in the table above refers to. (Your meter may also show short-term and momentary LUFS, which are 3-second and 400 ms windows — useful while mixing, but not what anyone's spec is written against.)

Why Mono and Stereo Have Different Targets

You'll see the podcast recommendation written as -16 LUFS for stereo, -19 LUFS for mono, and the 3 LU gap confuses people.

A mono file plays through both speakers or earbuds at full level, which makes it sound about 3 dB louder than the same content in a stereo file. So a mono file mastered to -19 LUFS and a stereo file at -16 LUFS sound equally loud on playback. If you master a mono podcast to -16 LUFS, it will actually play back louder than everyone else's shows.

Practical rule: master to -16 if your episode file is stereo (music beds, two-channel interviews), -19 if you export mono (single voice, smaller file).

True Peak: The Other Number That Matters

Loudness targets always come paired with a true peak ceiling, usually -1 dBTP.

True peak differs from the sample peak your DAW shows. When audio is converted back to an analog waveform — or encoded to a lossy format like MP3 or AAC — the reconstructed wave can swing between samples, exceeding the highest sample value by 0.5-1 dB or more. A file whose samples peak at -0.2 dBFS can clip audibly after encoding.

That's why platforms ask for -1 dBTP (Spotify suggests -2 dBTP if your show is quieter than -14 LUFS): it's headroom for the encoder. Use a limiter that's explicitly labeled "true peak," and set the ceiling at -1.0 to -1.5 dBTP.

What Loudness Normalization Does to Your Episode

Here's the part that changes how you should think about levels: most platforms adjust your loudness on playback whether you like it or not.

• Spotify measures your episode's integrated loudness and turns it down toward -14 LUFS if it's louder. By default it also turns up quieter shows (where limiting allows), and users can pick quiet/normal/loud playback environments (-19/-14/-11 LUFS).
• YouTube turns loud content down to about -14 LUFS. It never turns quiet content up — a -20 LUFS episode just plays 6 dB quieter than everything else on the platform.
• Apple Podcasts applies Sound Check-style normalization around -16 LUFS on modern devices, though listeners can toggle it.

Two consequences:

1. Mastering louder than the target buys you nothing. A podcast slammed to -10 LUFS gets turned down to -14 on Spotify — you paid for that loudness with crushed dynamics and got zero extra volume.
2. Mastering quieter than the target can genuinely hurt you, because normalization down is universal but normalization up is not. Which brings us to the real cost of quiet audio.

Why Quiet Podcasts Lose Listeners

Podcast listening is mostly mobile: commutes, gyms, kitchens, cars. Your episode competes with road noise at roughly 70 dB SPL and phone speakers with almost no headroom.

If your show sits at -22 LUFS while the previous show in someone's queue played at -16, yours starts 6 dB quieter — roughly half the perceived loudness. The listener's volume is already at maximum on the phone speaker over the dishwasher. They can't turn you up. Skipping to the next episode in the queue is one tap.

Inconsistency inside an episode is just as costly. If your intro music is 8 LU louder than your interview guest, listeners ride their volume control — and every manual adjustment is a moment of friction attached to your show. The fix isn't compression-to-death; it's hitting a consistent integrated target episode after episode, with your guest and host voices balanced within about 2 LU of each other before you measure the whole mix.

How to Hit the Target

The good news: unlike EQ or de-essing, loudness is a solved, mechanical problem. The process is called loudness normalization — measure the file's integrated LUFS, apply the exact gain needed to reach the target, and true-peak limit so nothing exceeds the ceiling.

The workflow:

1. Mix first. Balance voices, music, and transitions so nothing needs riding. Get host and guest within ~2 LU of each other.
2. Measure integrated LUFS on the full episode. Suppose you're at -21.5 LUFS aiming for -16: you need +5.5 dB.
3. Apply the gain, with a true-peak limiter at -1 dBTP catching whatever the boost pushes over. If the limiter is working hard (more than 2-3 dB of gain reduction on speech), your mix has isolated spikes worth fixing at the source — usually laughs, plosives, or a hot music bed.
4. Re-measure after limiting. Heavy limiting reduces loudness slightly; you want the final file within ±0.5 LU of target.
5. Encode last. Normalize the WAV, then encode to MP3/AAC — not the other way around.

You can do this in any DAW with a loudness meter, or skip the manual steps: ShiftAudioFormat has a normalize option that measures your file and hits your chosen LUFS target with true-peak limiting in one pass, and the Spotify Podcast preset bundles the -14 LUFS target, -1 dBTP ceiling, and encode settings together. It runs entirely in your browser — the episode never uploads to a server, and processing a one-hour episode takes about the time of a coffee refill.

One Target or Per-Platform Masters?

Should you export a -14 LUFS version for Spotify and a -16 for Apple? Almost never worth it.

Your podcast feed delivers one file to every platform, so you're choosing a single number anyway. And the platforms that want -14 (Spotify, YouTube) normalize on playback — a -16 LUFS file simply gets turned up ~2 dB where the limiting allows, or plays 2 dB quieter, which is barely perceptible. A -14 file on Apple gets turned down 2 dB by Sound Check. Either choice lands within 2 LU everywhere.

-16 LUFS stereo (-19 mono) is the sensible default for talk content: it's Apple's spec, it's the AES streaming recommendation for podcasts, and it leaves your dynamics a little more breathing room than -14. If your show is music-heavy and lives primarily on Spotify or YouTube, -14 is a defensible choice too. What actually matters is picking one number and hitting it within ±1 LU on every single episode — consistency is the whole game.

Quick Checklist

Before you publish an episode:

• Integrated loudness within 1 LU of your target (-16 LUFS stereo / -19 mono)
• True peak at or below -1 dBTP
• Host and guest voices within ~2 LU of each other
• Music beds sitting 12-18 LU under speech, not fighting it
• Same target as your last episode (your back catalog should match)
• Loudness measured on the final encoded file, not just the mix

Hit those six and your show plays back at the same comfortable level as the biggest podcasts in your listener's queue — which is exactly the point. Loudness done right is invisible; nobody ever compliments it, but everybody notices when it's wrong.