Podcast loudness standards: -16 LUFS, -19 LUFS and true peak, explained

Listeners rarely complain that a podcast is well mixed. They do notice when they have to turn it up for the intro, down for an ad, and up again for a quiet guest. Loudness standards exist so that doesn't happen, and so your show sits at a similar level to the one they played before it.

The short answer

• Stereo episodes: about -16 LUFS integrated.
• Mono episodes: about -19 LUFS integrated.
• True peak: no higher than -1 dBTP (many producers leave a little more room, such as -1.5 dBTP, for the encoder).

Apple's podcast guidance asks for around -16 LUFS with peaks under -1 dB, and the Audio Engineering Society's recommendation for streamed and podcast audio gives a range of -16 to -20 LUFS with the same peak ceiling. -16 sits at the loud end of that range, which suits listening on phones in noisy places.

What LUFS measures

LUFS (Loudness Units relative to Full Scale) measures how loud audio sounds over time, weighted for how human hearing works, as defined in the EBU R128 and ITU-R BS.1770 standards. "Integrated" loudness is the average over the whole episode, ignoring silence.

That's different from peak level, the loudest single instant. A recording can peak near 0 dB because of one laugh and still be quiet overall. Normalising to a peak, the way many editors do by default, doesn't make episodes consistent; normalising to LUFS does.

True peak (dBTP) estimates the peaks that appear between digital samples once the file is converted back to sound or encoded to MP3. Keeping it under -1 dBTP stops MP3 and AAC encoding from clipping.

Why mono is 3 dB lower

A mono file plays the same signal through both ears. Loudness meters measure a mono file as one channel, so at the same LUFS reading it sounds louder in headphones than a stereo file. Mastering mono about 3 dB lower (-19 LUFS) makes them sound alike, which is why the two numbers travel together.

If your show is one microphone per voice and nothing panned, a mono file at -19 LUFS is a sensible choice and half the file size of stereo.

How to check an episode

1. Measure the integrated loudness of the finished, edited episode (not the raw recording; edits change it).
2. Check the true peak.
3. Listen to the quietest speaker and the loudest moment on phone speakers and in headphones.

Free tools such as ffmpeg's loudnorm filter, and most audio editors, report integrated LUFS and true peak.

How to fix it

• Level the voices first. If one guest is much quieter, raise them (or compress the episode lightly) before normalising, otherwise the quiet voice stays quiet relative to the loud one.
• Normalise the whole episode to the target, with a limiter keeping true peaks under the ceiling. Two-pass normalisation measures the episode first and then applies one even adjustment, which avoids pumping.
• Trim the dead air. Long silence at the start and end doesn't change integrated loudness much, but it's the first thing listeners skip.

Where Podra fits

Podra masters every episode this way automatically: it trims the silence at the start and end, measures the episode, and normalises it to -16 LUFS for stereo or -19 LUFS for mono with true peaks under -1.5 dBTP, using ffmpeg's two-pass EBU R128 loudness normalisation. The episode page shows the loudness before and after, and you can try it on a sample without an account.