Normalize Audio Loudness: LUFS Targets Explained
Stop volume-jumping between tracks: normalize any recording to the right LUFS target for YouTube, podcasts, or broadcast — free, in your browser.
We’ve all been there: a playlist where one track blasts and the next whispers, or a video you upload that YouTube quietly turns down behind your back. The culprit is loudness mismatch. Recordings made at different volumes don’t match, and every platform, YouTube, Spotify, podcast apps, enforces its own loudness standard whether you like it or not.
Loudness normalization sets your audio to a standard target measured in LUFS (Loudness Units relative to Full Scale), which tracks how humans actually hear volume. Raw peak levels lie to you constantly; LUFS doesn’t. The Audio Normalizer measures your file and lands it on the right target in one click.
The three numbers that matter
Every platform picked a loudness level and normalizes everything toward it. Match the target and your audio plays back exactly as you mastered it. Miss it, and the platform steps in and does it for you.
- YouTube & Spotify: −14 LUFS. The streaming standard. Anything louder gets turned down automatically, so mastering hotter than −14 buys you nothing except squashed dynamics. Ask me how I learned that. Actually, don’t.
- Podcasts: −16 LUFS stereo, −19 mono. The podcasting convention, and every major host has converged here. Consistent loudness across episodes means listeners stop riding the volume knob in the car.
- Broadcast (EBU R128): −23 LUFS. European TV and radio. You only need this one if a broadcaster is actually asking for it.
The Normalizer has one-click presets for all three. Memorize nothing.
Normalizing in four clicks
- Open the Audio Normalizer and upload your recording. MP3, WAV, whatever the browser can decode.
- Pick the target: YouTube (−14 LUFS), podcast (−16 LUFS), broadcast (−23 LUFS), or type your own number if you’re feeling precise.
- Process. The tool measures integrated loudness with the BS.1770 algorithm, the same standard broadcasters use, and applies exactly the gain needed. Nothing more.
- Check the readout to confirm it landed on target, then download.
And that’s the whole trick: normalization is gain-only. No compression, no EQ, no change to your sound’s character. Just the level, set right.
Peak normalization is a trap
Most editors offer peak normalization, which cranks the loudest spike up to 0 dB and calls it a day. Sounds reasonable. It’s nearly useless.
Picture a whisper-quiet podcast with one loud cough in it. Peak normalization sees the cough, sets everything relative to the cough, and your podcast is still whisper-quiet. The feature did exactly what it promised and helped not at all.
Loudness normalization measures average perceived level across the whole file, gated, so silence doesn’t drag the average down, and adjusts that. Two files at −16 LUFS sound equally loud whether one’s a dense rock mix and the other’s sparse speech. When you’re matching material for the same platform, loudness wins every time. Peak normalization has exactly one job: squeezing the last dB out of a finished master. That’s it.
Mistakes I see constantly
- Normalizing a clipped recording. If it clipped on the way in, that harsh digital crunch, normalization can’t unbake it. The distortion is part of the file now. Re-record with lower input gain. I know that’s not what you wanted to hear.
- Normalizing before you’re done editing. Loudness goes last. Cut, mix, process, then set the final level once. Normalizing twice just stacks rounding error for no reason.
- Limiting the life out of it to hit −14. If your mix needs squashing to reach target without clipping, back off instead. Leave headroom, let the platform turn it down, keep your dynamics. Loud and flat loses to quiet and alive, every time.
- Ignoring true peak. After the loudness move, peaks need to sit below −1 dBTP or the MP3/AAC encode will clip them again. The tool handles this, just don’t fight it.
A workflow that just works
Loudness targets earn their keep as the last step of a repeatable routine:
- Podcast episode: edit and de-noise first, add light compression so quiet speech stays audible, then normalize to −16 LUFS stereo (−19 for mono). Export MP3 at 128 kbps. Voice doesn’t need 320, save the bytes.
- YouTube video: mix dialogue clearly on top of music and effects, then normalize the finished mix to −14 LUFS. YouTube leaves it alone, and your video matches everything else on the platform.
- Music single: finish the mix with peaks around −1 dBTP, normalize to −14 LUFS for streaming. Keep a louder un-normalized master only if a DJ specifically asks for club play, and eventually, one will.
Same pattern every time: creative work first, loudness last, one target per destination. Do that and you’ll never wonder why an upload sounds quiet again.
Related guides
Frequently asked questions
What exactly is LUFS?
The standard unit for perceived loudness, defined in ITU-R BS.1770. Unlike a peak meter, LUFS weights frequencies the way your ears do and averages over time, so the number predicts how loud something actually sounds. It’s why the whole industry switched.
Will YouTube turn down my video if it is louder than −14 LUFS?
Yes, YouTube normalizes everything toward roughly −14 LUFS, turning louder uploads down. (It barely turns quiet ones up, so don’t try gaming it from below.) Master to −14 yourself and your dynamics survive exactly as you intended, instead of being reshaped by an algorithm.
Does normalization reduce audio quality?
No. One clean gain change: no compression, no EQ, no artifacts beyond whatever format you export to. Export WAV and there’s zero generational loss, full stop.
Should podcasts be stereo or mono, and does it change the target?
Mono, in most cases, smaller files, and voices sit centered on every playback system from earbuds to smart speakers. Convention is −19 LUFS for mono versus −16 for stereo. The Normalizer’s podcast preset targets the stereo number; mono files naturally measure a touch lower, which is expected, not a bug.
My recording is too quiet even after normalizing — why?
Because normalization sets the average, not the punch. Huge dynamic swings, whispered asides, a distant mic, keep the average low even after the gain lift. For speech, gentle compression or manual leveling before normalizing is what actually evens things out.