How to Match Loudness Across Album Tracks: LUFS, Not Peak

In short. Tracks are matched by integrated loudness in LUFS, not by peak: two tracks with the same −1 dBFS peak can differ by 6–10 LU. Put the loudest track on the album at your target — for streaming usually around −14 LUFS — keep the others quieter by the few LU you intended, and hold True Peak at −1 dBTP or lower.

Below: why peak normalization fails, how to choose a target for an album, how to do it by hand, and how to do it online in Magic Master as a single batch.

Why album tracks sound uneven

Album tracks are often recorded and mixed on different days, in different moods, sometimes in different studios. Each one sounds fine on its own, but in sequence they do not: one jumps out, the next feels limp, and the listener reaches for the volume knob. Matching loudness solves exactly that: the tracks should sound like parts of one record.

Why peak normalization does not work

The usual first move is to "normalize" every track in an editor — raising each one until its highest sample sits at −1 or 0 dBFS. That does not match loudness.

A peak is the single highest moment in a file, such as one snare hit. Perceived loudness follows the average energy of the signal over time. A dense, compressed track and a dynamic live track with the same peak can differ by 6 to 10 LU: in the first nearly all the signal sits near the ceiling, in the second only rare hits touch it.

That is why loudness is measured in LUFS — units defined by ITU-R BS.1770, which weight the signal by the ear's sensitivity and average it over the whole track. Spotify, YouTube and other platforms normalize by LUFS. The scale is explained in "What is LUFS".

One target for the album, not for every track

If you bring every track on an album to −14 LUFS separately, the result is even but wrong: a quiet piano interlude gets blown up to the level of a dance single, and the album's intended dynamics disappear.

The right scheme is:

1. Choose a target for the loudest track. For streaming, around −14 LUFS integrated is a sensible reference; club music and "louder" genres sometimes aim higher.
2. Let the other tracks follow it. A ballad can be a few LU quieter, an interlude quieter still. What matters is not sameness but that the differences are intended rather than accidental.
3. Check the transitions. The end of one track and the start of the next should not produce a jump you did not plan.

This matches how the platforms behave: when an album is played in order, Spotify normalizes it as a whole and keeps the loudness differences between tracks. If those differences are intentional, they reach the listener.

True Peak: −1 dBTP or lower

The second rule is about peaks — true ones. True Peak accounts for inter-sample peaks: overshoots between digital samples that appear when the analogue signal is reconstructed and when audio is re-encoded to AAC, MP3 and Ogg. Keep every track at −1 dBTP or lower: that headroom protects against clipping after the platform's codec. Why it matters is covered in "What is True Peak".

How to match track loudness by hand

If you work in a DAW yourself:

1. Mix or master each track and measure its integrated loudness in LUFS across the whole track, start to finish.
2. Find the loudest one and bring it to the target with a limiter.
3. For the rest, decide how much quieter each should be and adjust the gain into the limiter.
4. Check each file's True Peak — −1 dBTP or lower.
5. Listen to the album in order, especially the transitions, and fix whatever jumps out.

It works, but it is slow, especially when the first listen means redoing several tracks.

How to match album loudness online in Magic Master

In the app it is one batch:

1. Upload all the tracks at once — up to 10 files at a time, up to 50 once you have bought tokens. WAV, FLAC, MP3 and AIFF, up to 800 MB in total.
2. Turn on Album mode in batch processing. In the new version of the interface it is on automatically for several tracks.
3. The service measures every track first and picks the anchor — the loudest one. It is mastered exactly to the target LUFS, and the others keep their original difference from it, going no more than 8 LU below the anchor.
4. One preset and one loudness target for the whole batch. That is what keeps the album coherent.
5. Download the tracks as they finish — they arrive one by one, so you do not wait for the whole album.

One important detail: Album mode keeps the differences that were in the input files. If the ballad came out of the mix louder than you wanted, the mode will not correct it — fix its level in the mix before uploading. And to be clear about the limits: the mode matches loudness, but it does not edit gaps or add crossfades, and it does not push the tracks towards one tone — reference mastering is for that. More on the album mastering page.

Each track in the batch is a separate master. The free allowance (a guest gets 3 a month, an account 1 a day) will not cover a whole album; with Telegram Stars a 10-token pack is 450 Stars. Other options are worked out in "How much does track mastering cost".

If you need to even out loudness within one track

Sometimes "even out the volume of a track" means something else: a quiet verse and a loud chorus in one recording, or a voice that jumps up and down. Mastering smooths this partly with compression and a limiter, but big swings inside a track are better fixed earlier, with volume automation in the mix. Mastering sets the overall loudness and balance; it does not rewrite the mix.

Check the result with an analyzer

After mastering, run the finished files through the free LUFS analyzer. It shows the integrated loudness and True Peak of each track. A good result looks like this: the loudest track near the target, the others below it by the amount you intended, and True Peak nowhere above −1 dBTP. Finally, listen to the album in order, first track to last: the numbers confirm, but your ears decide.