Loudness for Podcasts, YouTube and Audiobooks (-14 LUFS Explained)
Updated September 3, 2026
TL;DR: LUFS measures perceived loudness over time; true peak measures the highest instantaneous level; RMS is an older average. Streaming platforms normalise to around -14 LUFS, so mastering there means your audio is neither turned down nor left quiet. Audiobook retailers specify RMS (−23 to −18 dBFS) and peak (≤ −3 dB) instead. LeapFun exports at -14 LUFS / TP -2 / 48 kHz by default, and has an Audiobook preset for the retailer spec.
The three numbers
| Measure | What it captures | Typical target |
|---|---|---|
| Integrated loudness (LUFS) | Perceived loudness across the whole file, weighted to how ears hear | -14 LUFS for YouTube, Spotify, TikTok-style platforms |
| True peak (dBTP) | Highest instantaneous level including inter-sample peaks | -1 to -2 dBTP for streaming; audiobook spec uses peak ≤ -3 dB |
| RMS (dBFS) | Average signal power, no perceptual weighting | -23 to -18 dBFS for the common audiobook spec |
Why -14 LUFS
Platforms normalise: louder files are turned down, quieter files may be turned up (or not). Mastering at the platform target means the audio arrives at the level it will be played, with the dynamics you chose. For speech, a loudness range around 6 LU keeps narration consistent without sounding squashed.
Why audiobooks use RMS and peak
Audiobook technical specs predate LUFS and were written for spoken word in one voice: RMS between −23 and −18 dBFS, peaks no higher than −3 dB, 44.1 kHz, 192 kbps CBR MP3, files under 120 minutes, quiet room tone at the head and tail. For speech, RMS and LUFS land close, but the peak ceiling is stricter and the sample rate differs — which is why a streaming master is not automatically a retailer-ready file.
What LeapFun does
- Standard preset (default everywhere): -14 LUFS integrated, -2 dBTP, loudness range ≈ 6 LU, 48 kHz. Drops into a video or podcast edit at platform level.
- Audiobook preset (Studio export): masters to a -20 LUFS starting point with a -3.5 dBTP ceiling at 44.1 kHz — the extra headroom exists because MP3 encoding raises peaks — then measures the encoded file against the retailer spec and shows a report.
- Music beds are mixed before mastering, so the mix, not just the voice, is normalised.
- If mastering can't run (a near-silent file, for instance) the export still completes and the report says it was not normalised — no silent failure.
Practical rules
- Don't re-normalise in your editor unless a platform demands a different target.
- If the voice seems quiet next to music, lower the music.
- Check loudness on the encoded file, not the WAV — encoding changes peaks.
- For retailers, use the Audiobook preset and read the report; then check their policy on AI narration separately.
Frequently asked questions
Is -14 LUFS right for podcasts?
It's the common target for major podcast platforms and works well for spoken word. Some networks specify -16 or -18; if yours does, master to that.
Why does the audiobook preset start at -20 LUFS rather than an RMS value?
Mastering tools work in LUFS; for speech, -20 LUFS lands the encoded file's RMS in the middle of the -23 to -18 dBFS window. The pass/fail is measured on the final MP3, not assumed from the setting.
Does mastering change how the voice sounds?
Level and peak control only — no EQ or compression beyond the true-peak limiter. The timbre is the voice's.
Can I turn mastering off?
Not per export in the app; it's on by default so files land at platform level. If you need raw output for a custom chain, contact support.