Level 3Lesson 16 of 248 min

Loudness and LUFS

Two episodes with the same peak can sound twice as loud as each other. What the listener actually hears is loudness, and it's measured in LUFS. Here you'll learn what number to aim for, why, and how to get there without ruining the voice.

Peak versus loudness

So far you've been looking at the peak: the highest point of the waveform in dBFS. It stops you from clipping, but it tells you little about how loud something sounds. A sharp hit lasting an instant can reach −1 dBFS, and yet a continuous, compressed voice peaking at −6 dBFS will sound much louder.

Loudness is perceived volume: it takes into account how long the signal lasts and which frequencies the ear is most sensitive to. There is a standard unit to measure it, LUFS (Loudness Units relative to Full Scale). Like dBFS, the values are negative: −14 LUFS sounds louder than −16 LUFS.

1 LU is 1 dB

One loudness unit (LU) equals one decibel. If your episode is at −19 LUFS and you want it at −16, you need to add 3 dB of gain.

The readings you'll see on a meter

Reading	What it shows	What you use it for
Integrated	The average loudness of the whole episode	This is the number you need to hit
Short-term	Loudness over the last 3 seconds	Spotting a segment or guest that sounds quieter
Momentary	Loudness over the last 400 ms	Instant feedback; useful when mixing live
True peak (true peak, dBTP)	The estimated peak after conversion to analog or MP3	Keep it below `−1 dBTP`
Loudness range (LRA)	How much the loudness varies across the episode	If it's very high, you need more compression or some segments are uneven

True peak deserves an explanation: when digital audio is converted (and when it's encoded to MP3 or AAC), peaks can appear between samples that go above what a normal peak meter shows. That's why you leave a margin and measure in dBTP.

What target to aim for

There's no single law for podcasts, but there is a recommendation that's widely used across the industry:

Format	Integrated loudness	Maximum true peak
Stereo podcast	`−16 LUFS`	`−1 dBTP`
Mono podcast	`−19 LUFS`	`−1 dBTP`

Why 3 LU less for mono? Because when a mono file plays through two speakers or two earbuds, the same signal comes out of both sides and is perceived as louder. At −19 LUFS, mono sounds practically the same as −16 LUFS stereo. Careful: some meters already compensate for this, so check what your software shows.

Use the simulator above to move the output gain and see how the integrated LUFS and the true peak change relative to the −16 LUFS target.

Platforms normalize

Many platforms adjust the volume of what they play so everything sounds similar. Spotify and YouTube, for example, work with references around −14 LUFS, and other apps use different values. These criteria change over time and with user settings, so don't chase each platform's number.

If you deliver much louder than their reference, they'll turn you down: you'll have squashed the voice for nothing.
If you deliver much quieter, they might turn you up… or they might not (some don't if it would clip the peaks), and you'll sound weak next to everything else.
Many podcast apps don't normalize at all: what you deliver is what people hear.

One master for everything

For most projects, −16 LUFS stereo with the peak below −1 dBTP is an excellent compromise: it sounds good in apps that don't normalize and is barely touched in those that do.

How to measure and adjust

Finish the mix first: cleanup, EQ, compression and music. Loudness is adjusted at the end.
Measure the whole episode with a loudness meter. The integrated reading is only reliable if it covers all the audio.
Correct with gain: if you're at −18 LUFS and want −16, add 2 dB.
Control the peak with a limiter at the end, with the ceiling at −1 dBTP.
Measure again on the exported file, not just the project.

Many programs also have a loudness normalization effect that brings the whole file to a LUFS value in one go. It's convenient, but it's no substitute for compression: if the voice has a lot of dynamics, turning it up will make the peaks slam into the ceiling.

Not all normalization is the same

Peak normalization brings the highest peak to a value (for example, −1 dBFS) and guarantees nothing about loudness. Loudness normalization works with LUFS. For a podcast master, the second one is what you want.

Where to find a meter

You don't need anything expensive. Audacity, DaVinci Resolve and most audio programs include a loudness meter or an effect that analyzes the file and gives you the integrated LUFS and true peak. There are also free plugin meters. What matters is always using the same one so you can compare between episodes.

Common mistakes

Measuring only a section: the opening with the theme music can give a very different value from the whole episode.
Reaching −16 LUFS with the limiter alone: if the limiter is taking off 6 dB, the voice will sound squashed. Compress first.
Forgetting mono: exporting in mono a master made for −16 LUFS stereo without checking how your meter reads it.
Uneven guests: a good integrated value can hide one person sitting 6 dB below another. Even out each track before mastering.

In short

Peak keeps you from clipping; LUFS tells you how loud it sounds.
Usual target: −16 LUFS stereo or −19 LUFS mono, with the true peak below −1 dBTP.
Platforms normalize, but not all of them and not the same way: deliver a balanced master.
Measure the whole episode at the end, then measure the exported file again.

Your turn: measure your favorite podcast

Compare the loudness of a professional episode with your latest episode and bring yours to the target.

Use a loudness meter on a long section of a podcast you like and note down the LUFS and true peak.
Measure your whole episode and note the difference.
Correct with gain and a limiter until you hit −16 LUFS and −1 dBTP, export and measure again.

I've done it

Frequently asked questions

What LUFS should a podcast be?

The most widely used recommendation is −16 LUFS integrated for stereo and −19 LUFS for mono, with the true peak below −1 dBTP.

What is true peak?

It's the estimated real peak of the signal, including the peaks that appear between samples when the audio is converted or encoded. It's measured in dBTP, and for podcasts the recommendation is not to go above −1 dBTP.

If Spotify normalizes to −14 LUFS, should I master to −14?

There's no need. At −16 LUFS the change it applies will be small, and your episode will also sound good in apps that don't normalize.