Why podcast loudness matters more than you think
Every major platform measures the loudness of your episode and normalizes playback toward its own target. If your episode comes in hot, it gets turned down — and any limiting you used to get there stays baked in as distortion, now without the loudness payoff. If it comes in low, listeners reach for the volume knob, and everything you didn't want them to hear — room tone, breaths, preamp hiss — comes up with your voice.
The result is the same in both directions: your episode sounds different from the one that played before it, and different from the one that plays after. Listeners don't diagnose that as a loudness problem. They just register it as "this show sounds off."
Why −16 LUFS, not −14
Spotify and YouTube both normalize toward −14 LUFS, so −14 looks like the obvious target. It isn't — and the reason is spoken word.
The widely cited Apple Podcasts guideline is −16 LUFS for stereo. That's not Apple being conservative; it reflects how speech behaves. Music at −14 LUFS is dense — the energy fills the spectrum continuously. Speech at −14 LUFS has to be compressed noticeably harder to hold that level through pauses and phrasing, and that's exactly when voices start sounding pinched and fatiguing on earbuds.
Master spoken word at −16 LUFS and every platform handles it gracefully: Apple plays it as intended, and Spotify's normalization brings quieter content up toward −14 on platforms that want it louder. You lose nothing and keep the natural dynamics that make a voice pleasant for forty minutes, not just forty seconds.
True peak: the −1 dBTP rule
Your DAW's peak meter reads the samples. Your listener's phone plays a reconstructed waveform — and between two samples, that waveform can swing higher than either of them. These inter-sample peaks are invisible on a sample-peak meter and become real, audible clipping after the AAC or MP3 encode every platform applies.
That's why the spec is −1 dBTP (decibels True Peak), not −1 dBFS. A limiter with true-peak detection set to −1.0 or lower leaves the codec room to work. If your episode measures above −1 dBTP, it may sound clean on your machine and crackle on your audience's.
Loudness range: dynamics are a delivery decision
Integrated LUFS tells you how loud the episode is overall. LRA (loudness range) tells you how much it moves. For spoken word, roughly 5–11 LU is the comfortable zone: enough movement that a voice sounds like a person, not so much that a listener in a car keeps riding the volume.
Below that range, the audio has usually been compressed hard enough that breaths and room pumping become part of the sound. Above it, the quiet passages disappear under road noise and the loud ones startle. Neither shows up on a level meter — both show up in whether people finish the episode.
A number is not a diagnosis
Any free LUFS meter can tell you your episode is −12.3 LUFS. What it can't tell you is what that means: that you're 3.7 LU hot for Apple, that the platform will turn you down, that the limiting used to get there is now pure cost, and what specifically to change in your chain before the next episode.
That's the difference this tool is built around. The measurements follow ITU-R BS.1770-4 — the same standard broadcast delivery is checked against — and the interpretation comes from twenty years of mixing for broadcast, where a delivery that misses spec gets sent back.
If you'd rather not manage any of this per episode: PodMaster is our mastering pipeline built on the same broadcast targets — it lands every episode at spec, consistently, so the checker becomes a confirmation instead of a to-do list.
For the full walkthrough of targets and how to hit them in your own chain, read our guide: How loud should a podcast be?