Podcast loudness: -16 LUFS, -14 LUFS or -18?

For a podcast, aim for -16 LUFS integrated loudness with true peak at or below -1 dB. That is the figure Apple Podcasts publishes, and it plays well everywhere else too. The -14 LUFS number you see in forums is Spotify’s music target, and -18 LUFS is an AES recommendation for speech. This post explains what each number means, where it comes from, and the order to process a file so it lands on target.

What the -16, -14 and -18 numbers actually are

Each figure comes from a different place, and only one of them is a published podcast requirement.

Platform or body	Published loudness	Peak guidance	What it applies to
Apple Podcasts	Around -16 LKFS, ±1 dB	True peak not above -1 dBFS	Podcasts, official requirement
Spotify	-14 LUFS (Normal); -11 Loud, -19 Quiet	Below -1 dBTP; below -2 dBTP if louder than -14 LUFS	Music tracks and albums
Spotify podcasts	No figure on the podcast specification page	Not stated	Podcasts
YouTube	No official figure found	Not stated	Third-party measurements put it near -14 LUFS
AES TD1008 (2021)	-18 LUFS speech, -16 LUFS music	Not covered here	Streaming distribution recommendation
EBU R 128	-23 LUFS	-1 dBTP	European broadcast

A few notes on the table:

• Apple says it recommends that audio is “preconditioned” so the overall loudness stays around -16 dB LKFS with a ±1 dB tolerance, and that true peak does not exceed -1 dB FS. Both values are calculated per ITU-R BS.1770-5. Apple also says this preconditioning has to happen before encoding.
• Spotify’s page is written for artists. It says Spotify adjusts tracks to -14 dB LUFS per ITU 1770, applies positive gain to quieter masters, and uses a limiter on the Loud setting. The page does not mention podcasts. Spotify’s podcast specification page describes file format and metadata requirements and does not give a loudness number.
• YouTube has no official help page with a loudness target that I could find. Production Advice measured playback near -14 LUFS and reports that YouTube turns loud videos down and does not turn quiet ones up. Treat -14 for YouTube as an observed behavior, not a spec.
• AES TD1008 is a technical document from the Audio Engineering Society. According to Production Advice’s summary, it recommends -18 LUFS for speech-only and mixed “assorted” content, and -16 LUFS for music.

LUFS and LKFS are the same unit

LKFS is the name ITU-R BS.1770 uses. LUFS is the name the EBU uses. EBU R 128 states plainly that LUFS is equivalent to LKFS. Apple’s page says -16 LKFS, your meter probably says -16 LUFS, and they mean the same thing.

Both measure loudness relative to digital full scale, as defined in ITU-R BS.1770. That is why a podcast and a song can peak at the same level and still sound very different in loudness. Peak level tells you about the tallest spike. LUFS tells you about how loud the program feels over time.

Integrated, short-term and momentary loudness

A loudness meter shows several readings. EBU Tech 3341 defines them:

• Momentary loudness uses a sliding 0.4 s window. It is not gated.
• Short-term loudness uses a sliding 3 s window. It is not gated.
• Integrated loudness covers the whole program, and it is gated as described in ITU-R BS.1770.

Platform targets refer to integrated loudness. A host who gets excited will push short-term readings well above -16 for a few seconds, and that is fine. What counts is the integrated figure for the whole episode.

Use short-term readings to spot problems: a guest who sits 6 dB below the host, an ad read that jumps out, or a music bed that is too hot. Fix those with gain or compression on the affected section, then measure the integrated value again.

True peak and why the limit is -1 dB

Sample peak is the highest sample value in the file. True peak estimates the highest level the waveform reaches between samples once it is converted back to analog or decoded. A file can show sample peaks below 0 dBFS and still have true peaks above it.

Lossy encoding can make this worse. Apple warns that compression may clip a signal that does not respect the true-peak value. That is why Apple asks for true peak at or below -1 dBFS, EBU R 128 sets -1 dBTP, and Spotify’s music page suggests below -1 dBTP, or below -2 dBTP for masters louder than -14 LUFS. A true-peak limiter set to -1 dBTP is the simplest way to stay under all of these.

Why removing silence changes integrated loudness

You might expect silence to drag the average down a lot. It mostly does not, because of gating. EBU Tech 3341 summarizes the BS.1770 method:

1. The audio is measured in 400 ms blocks that overlap by 75%.
2. Blocks below an absolute gate of -70 LUFS are discarded.
3. A relative gate is set 10 LU below the loudness of the blocks that remain, and blocks under it are discarded too.
4. The integrated value is computed from the blocks that pass both gates.

Digital silence falls under the -70 LUFS gate. Quiet room tone in a pause usually falls under the relative gate. Neither counts.

The blocks that do count are the ones at the edges of pauses. A 400 ms block that holds the end of a word and the start of a pause passes the gate at a lower level than a block of full speech. Trailing words, breaths and soft starts do the same. When you cut pauses, you remove many of those partial blocks, and the integrated reading usually moves a little. The size of the shift depends on the recording, so measure rather than guess.

The practical point is simple: if you normalize first and cut silence afterward, the file you publish is no longer the file you measured.

Order of operations: cut, normalize, limit

Process a podcast in this order:

1. Edit first. Remove dead air, long pauses, false starts and filler words. Set levels between speakers and fix any section that is far too loud or quiet.
2. Measure integrated loudness of the edited file.
3. Normalize the whole file with gain so the integrated reading is -16 LUFS.
4. Limit true peak to -1 dBTP with a true-peak limiter.
5. Measure again. If the limiter had to work hard, the integrated value can drop. Adjust gain and repeat.
6. Encode to MP3 or AAC, then spot-check the encoded file.

In Audacity, loudness normalization is under Effect > Volume and Compression > Loudness Normalization. Its default target is -23 LUFS, the EBU broadcast figure, so change it to -16 for a podcast. The manual does not say the effect controls peaks, so check true peak afterward. The manual also describes a Treat mono as dual-mono option that is on by default, which matters if your show is mono and your meter treats mono differently.

Choosing -16, -14 or -18 for your show

For an audio podcast, -16 LUFS integrated with true peak at or below -1 dB is the safe default. It matches Apple’s published requirement, and it leaves room for a limiter to work without squashing your voice.

Consider -18 LUFS if your show is quiet and intimate, like a narrative piece or a sleep podcast, and you prefer more dynamic range. In any app that does not raise quiet audio, it will play about 2 dB below shows mastered to -16.

For a video podcast on YouTube, there is no official number to aim at. Based on Production Advice’s measurements, a -16 LUFS mix will not be turned down there, and it will not be turned up either, so it will play a little below content that hits about -14. You can publish the same -16 LUFS master everywhere and accept that difference. If you want parity with loud YouTube content, you can make a separate -14 LUFS mix, but keep true peak at or below -1 dBTP.

File formats Apple lists

Loudness is only half of a clean delivery. Apple’s audio requirements page lists these formats:

Delivery	Formats	Notes from Apple
Apple Podcasts Connect (subscriber audio)	WAV, FLAC or MP3	Mono not accepted for WAV or FLAC; use two identical channels. Minimum 44.1 kHz.
RSS feed	MP3 or AAC	Apple strongly recommends AAC, in the MP4 container over ADTS
RSS, mono at 44.1/48 kHz	64–128 kbps	Same bit rates for AAC and MP3
RSS, stereo at 44.1/48 kHz	128–256 kbps	Same bit rates for AAC and MP3

Apple also notes that for the same bit rate, AAC gives better audio quality than MP3.

Mistakes that throw the number off

• Normalizing before the edit. Cutting pauses, intros or ad breaks afterward changes integrated loudness.
• Using peak normalization. Normalizing peaks to -1 dB says nothing about LUFS. A spoken track can peak at -1 and still measure far below -16.
• Skipping the true-peak check after encoding. MP3 and AAC can overshoot. Check the encoded file.
• Measuring one section. Integrated loudness is for the whole episode. A 2-minute sample can read differently from the full show.
• Pushing quiet audio up by gain alone. If the voice sits very low, gain to -16 will push peaks into the limiter. Use gentle compression first.

Doing it on a phone or in the browser

If your recording starts with long gaps, cut them before you think about loudness. The free silence remover on this site runs in your browser, shows every gap it finds, and lets you download the cleaned audio as WAV, which you can then normalize in your editor. Recordings up to 3 minutes are free and the file never leaves your device. For tuning the cut itself, see silence threshold settings and what dead air is.

On iPhone or iPad, QuellCut handles the order for you: it removes silence and filler words, lets you review each cut on a timeline, and normalizes the audio to -16 LUFS, all on the device. If your show is recorded on remote calls, the sister guide on removing pauses from Zoom podcast recordings covers that workflow.

Questions

Is -16 LUFS the same as -16 LKFS?

Yes. The EBU states that LUFS is equivalent to LKFS, the unit used in ITU-R BS.1770. Apple writes LKFS, most meters and plugins write LUFS.

Should I master my podcast to -14 LUFS for Spotify?

Spotify's -14 LUFS figure is on its music page, and its podcast specification page does not state a loudness target. A podcast at -16 LUFS with true peak at or below -1 dB is a safe file for every major app.

What does AES recommend for speech?

According to Production Advice's summary of AES TD1008 (2021), the recommended distribution loudness for speech-only content is -18 LUFS, and -16 LUFS for music.

Does YouTube turn quiet videos up?

YouTube has not published its loudness rules. Production Advice's measurements found that YouTube turns loud videos down and does not turn quieter ones up.

Do I need to normalize again after cutting pauses?

Yes. Removing pauses changes what the meter averages over, so measure and normalize the edited file, not the raw recording.

My podcast is mono. Does that change the reading?

It can. Audacity's Loudness Normalization has a Treat mono as dual-mono option, on by default, because some meters read mono 3 LU lower. Check that your meter and your normalizer handle mono the same way.

Sources: Apple Podcasts for Creators: Audio requirements, Spotify for Artists: Loudness normalization, EBU R 128 (2023), EBU Tech 3341: Loudness metering, EBU Mode, Production Advice: AES TD1008 streaming loudness, Production Advice: YouTube Stats for nerds.