How loud should a podcast be?

Someone emails you that your podcast is too quiet in the car. You turn it up in your editor, export again, and now a listener with earbuds says the intro blew their ears out. Loudness feels like it should be the easy part of audio, and it refuses to be.

The fix is knowing exactly two numbers: your integrated loudness target and your true peak ceiling. For most spoken-word content those should be around -16 LUFS and -1 dBTP. Here is what those mean and where they come from.

One note before the numbers: the word podcast is in the title because that is how people search, but everything here applies to any spoken audio. YouTube videos, course content, recorded meetings and webinars, audiobook narration, conference talks. If humans talk in your file, these are your targets too.

What LUFS actually measures

LUFS stands for Loudness Units relative to Full Scale. Unlike a peak meter, which only sees the single highest sample, LUFS measures how loud a recording feelsover time, using a model of human hearing standardized in ITU-R BS.1770. When you see “integrated LUFS,” that is the average loudness of the whole file.

The scale is negative because 0 is digital maximum. A typical raw voice memo might sit around -26 LUFS. Commercial podcasts mostly play back between -14 and -16. The difference between those two, roughly 10 dB, is exactly the “why is my show so quiet” email.

What the platforms do with your file

Every major platform now normalizes loudness, meaning they turn your episode up or down toward their own reference level:

Two practical consequences. First, mastering to -16 LUFS means every platform either leaves you alone or makes a small, harmless adjustment. Second, and this is the part people miss: YouTube will not rescue a quiet upload. If you upload at -24 LUFS, viewers hear -24 LUFS until they reach for the volume.

Why not just make it as loud as possible?

Because normalization turns loudness wars into pure loss. If you crush your podcast to -10 LUFS, Spotify simply turns it down to -14, and what remains is the crushing: flattened dynamics, pumping artifacts, and listener fatigue, at the same playback volume as everyone else. Loudness maximization made sense before normalization existed. In 2026 it is self-harm.

Music is a slightly different conversation, with targets closer to -14 and more taste involved. For speech, the consensus around -16 is boring and correct.

The second number: true peak

The -1 dBTP ceiling exists because of what happens after mastering. When your file is encoded to MP3 or AAC for delivery, the waveform gets reconstructed slightly differently, and peaks can land between your original samples, higher than any sample you exported. Leave no headroom and those reconstructed peaks clip. We wrote a separate breakdown of this in why audio clips after mastering.

A workable checklist

• Record with peaks around -12 to -6 dB. Headroom is free at this stage.
• Master or normalize the final episode to about -16 LUFS integrated.
• Keep true peaks at or below -1 dBTP.
• Check the quietest passage, not just the average. If a soft-spoken guest sits 8 dB under the host, no single number fixes that; you need compression or per-track gain first.

Any decent DAW can measure LUFS these days, and plenty of free meters exist. FixAudio does this step automatically: it measures your file, masters to target, verifies the result, and keeps true peaks under the ceiling. That last verification pass matters more than the tool you pick. Loudness is one of the few parts of audio where you can simply check your work with a meter, so check it.