How to Normalize Audio to -14 LUFS

Audio normalization is how you make tracks play at the same perceived volume — and how you stop Spotify and YouTube from turning your masters down. This guide explains loudness normalization to -14 LUFS step by step: what it is, why peak normalization isn't enough, platform targets, and how to do it free in your browser.

If you've ever made a playlist where one song blasts and the next whispers, you've felt the problem normalization solves. And if you master music for streaming, normalization isn't optional anymore — Spotify, YouTube, and TikTok all normalize playback to around -14 LUFS whether you like it or not. Understanding how to normalize audio correctly is the difference between a master that translates and one that gets quietly turned down.

What is audio normalization?

Audio normalization means adjusting a file's gain so it hits a defined loudness target. The key word is loudness — modern normalization targets perceived loudness (LUFS), not electrical level. The process is simple: measure the integrated loudness, compute the difference to the target, apply that gain to the whole file.

Example: your mix measures -18.5 LUFS integrated. Target is -14 LUFS. Normalization applies +4.5 dB of gain. Done — transparent, no compression, no EQ, just a volume change calculated from a perceptual measurement. (New to LUFS? Start with What is LUFS?.)

This is fundamentally different from the old "normalize" button in audio editors, which brings us to the most important distinction in this guide.

Peak normalization vs loudness normalization

Peak normalization is nearly useless for delivery. A heavily limited EDM drop and a dynamic jazz trio can both peak at exactly 0 dBFS and differ by 8 LU in perceived loudness. Peak tells you about headroom; LUFS tells you about loudness. When someone says "normalize my MP3 volume" or "make these audio files the same volume," they mean loudness normalization — even if they don't know the term.

Why -14 LUFS? Platform loudness targets

-14 LUFS integrated is the de facto streaming target because Spotify (Normal mode, the default) and YouTube both normalize to approximately -14. If you deliver at -14, platforms leave your gain alone — no surprise turn-downs, no positive-gain limiting artifacts.

You don't have to hit -14 exactly — Spotify tolerates a range, and many engineers master to -12 or even -10 for punch, accepting the turn-down. What you should not do is master to -7 LUFS thinking "louder wins." It doesn't; the platform just turns it down 7 dB, and you've crushed your dynamics for nothing. Read the Spotify -14 guide and why -14 is the new 0 dB for the full story.

How loudness normalization works (4 steps)

1. Measure. Compute integrated LUFS over the full program with K-weighting and EBU R128 gating (-70 absolute, -10 relative). This is the only honest number — a 10-second snippet won't do.
2. Compute gain. gain_dB = target - measured. Measured -18.5, target -14 → +4.5 dB.
3. Apply gain in 32-bit float (or higher). This is bit-transparent apart from the level shift — no dynamics processing involved.
4. Check true-peak. Raising gain raises peaks too. If true-peak now exceeds -1 dBTP, you need a true-peak limiter pass or a lower target. Never ship a normalized file that clips inter-sample peaks — see True Peak.

Measured integrated:  -19.2 LUFS
Target:               -14.0 LUFS
Gain to apply:        +5.2 dB
True-peak before:     -6.3 dBTP
True-peak after:      -1.1 dBTP  → safe (under -1.0)
Result: file now averages -14 LUFS, peaks safe. Done.

How to normalize audio: step by step

You can normalize audio free in your browser with JSM Loudness — no upload, no install:

1. Open the Analyzer and drop your audio file.
2. Read the integrated LUFS value (and note true-peak).
3. Use the gain control to apply target − measured dB. For Spotify/YouTube, target -14.
4. Re-check: confirm integrated reads ≈ -14 LUFS and true-peak stays under -1 dBTP.
5. Export. If you need a different format too, run it through the audio converter after normalizing — always normalize the best-quality source first.

For a deeper understanding of what you're measuring, read how to measure audio loudness — it covers integrated vs short-term vs momentary and how to read a LUFS meter properly.

Normalizing for different destinations

Music for streaming: -14 LUFS integrated, true-peak ≤ -1 dBTP. Master with dynamics intact; let the platform handle playback level.

Podcasts: -16 LUFS stereo (-19 mono) is the podcast standard — speech is denser than music, so it needs a lower target to avoid listener fatigue. Keep LRA modest; wildly dynamic speech forces listeners to ride the volume.

Playlists / compilations: normalize every track to the same target. For albums, consider album-gain: compute one gain from the whole album's loudness so inter-track dynamics (the quiet ballad after the loud single) are preserved.

Video / broadcast: -23 LUFS per EBU R128 (or -24 for ATSC A/85 in the US). Different world, different spec — see EBU R128 in 5 minutes.

Common normalization mistakes

• Normalizing to 0 dB peak and calling it "normalized." Your tracks still won't match in loudness. Peak ≠ loudness.
• Normalizing each album track independently. Destroys intentional dynamics between songs. Use album gain.
• Ignoring true-peak. +4 dB of normalization gain on a hot master = inter-sample clipping. Always check dBTP after.
• Normalizing a lossy file then re-encoding. Transcoding stacks artifacts. Normalize the WAV/FLAC, then encode once.
• Chasing exactly -14.0. ±1 LU is inaudible and within every platform's tolerance. Don't crush dynamics to hit a number.

Frequently asked questions

Should I normalize audio to -14 LUFS or 0 dB peak?

Normalize to -14 LUFS integrated if your target is Spotify or YouTube — that matches their playback normalization, so your track won't be turned down unexpectedly. Peak-normalizing to 0 dB only sets the loudest sample to full scale; it says nothing about perceived loudness, so two 0 dB-peak tracks can still sound 6 dB apart. Use peak normalization for editing headroom, loudness normalization for delivery.

Will Spotify turn down my track if it's louder than -14 LUFS?

Yes. Spotify's default Normal setting applies gain to bring tracks to -14 LUFS integrated. A master at -8 LUFS gets turned down 6 dB — you lose nothing except the loudness-war advantage you thought you had. A master at -18 LUFS gets turned up 4 dB, but Spotify limits positive gain to avoid clipping, so quiet masters may stay quieter than intended. Mastering near -14 LUFS integrated with true-peak under -1 dBTP is the predictable path.

Does normalizing audio reduce quality?

Loudness normalization is a simple gain change — mathematically transparent when done in 32-bit float or higher. It doesn't compress, EQ, or otherwise alter the sound; it just turns the volume up or down. Quality loss only enters if you normalize a lossy file and re-encode it (transcoding), or if you push true-peak above 0 dBFS and clip. Normalize from WAV/FLAC and watch true-peak.

How do I make multiple audio files the same volume?

Measure the integrated LUFS of each file, then apply per-file gain so they all land on the same target (e.g. -14 LUFS for a playlist, -16 LUFS for podcast episodes). Don't just peak-normalize — files with identical peak levels can differ wildly in perceived loudness. For albums, normalize to the loudest track's gain or use album-gain so inter-track dynamics are preserved.