Most people learning to master their own music run into the same problem.
The mix sounds good while they are working on it.
They export it, it gets louder, and then it sounds worse. Not worse in an obvious way.
Just a little flat. A little tired. Less exciting than the mix was before.
This almost always comes down to the limiter. Specifically, to how much the limiter is being pushed and what it is doing to the transients in the process.
This post is part of the Complete Guide to Audio Mastering.
What a Limiter Does
A limiter is a compressor with a very high ratio, typically 20:1 or higher, sometimes described as infinity:1.
At that ratio, the signal essentially cannot exceed the threshold by any significant amount. Every peak that hits the threshold gets caught and brought down to the ceiling level you have set.
In mastering, the limiter sits at the end of the chain with the output ceiling set to -1 dBTP (true peak).
You then push the input level until the output reaches your target LUFS.
The gain reduction you see on the limiter meter is the difference between the original peaks in your mix and the ceiling you have set.
A limiter with 1 to 3 dB of gain reduction is doing its job correctly and probably making the master better.
A limiter with 8 to 10 dB of gain reduction is destroying the transients and likely making the master worse, even if the loudness meter reads exactly where you wanted.
Choosing the Right Limiter
Not all limiters sound the same. This is one of the most important decisions in mastering because the limiter affects more of the signal than any other processor. Every peak in the mix passes through it.
The most important quality to look for in a mastering limiter is transparency at reasonable gain reduction amounts.
You want a limiter that can handle 2 to 3 dB of gain reduction without the transients changing character. Some limiters do this better than others.
Well-regarded mastering limiters include the FabFilter Pro-L 2, Sonnox Oxford Limiter, Nugen Audio ISL, and the Weiss DS1-MK3. The iZotope Ozone limiter is also widely used, particularly the IRC algorithm settings.
The stock limiters in most DAWs are generally not the best choice for mastering, though they can work in a pinch at light gain reduction amounts.
True Peak vs Sample Peak
Set your output ceiling to -1 dBTP (true peak), not -1 dBFS (sample peak).
These are not the same measurement. Sample peak shows you the highest sample value in the digital file.
True peak measures the actual signal level after digital-to-analog conversion, which can exceed the sample peak level due to inter-sample peaks.
If you set your ceiling at -1 dBFS and your limiter is only measuring sample peaks, the true peak level may still be exceeding 0 dBFS, which causes distortion when the file is played back.
A true peak limiter mode on your limiter plugin handles this correctly.
See What is LUFS? for the full explanation of loudness measurement and true peak.
Loudness and Limiting: The Right Relationship
The loudness target you are trying to reach should determine how much limiting is appropriate. If your mix is already dynamic and well-controlled, reaching -14 LUFS integrated will require very little limiting.
If your mix has been compressed heavily during mixing and already has a high average level, the limiter needs to do less to reach the same LUFS target.
The problem arises when engineers try to push a dynamic mix to a loudness level it was not made for.
A mix that sounds best at -14 LUFS will sound worse at -9 LUFS because getting there requires so much limiting that the transients are destroyed.
The right decision is to master it at the loudness level that sounds best, not the loudness level that is technically possible.
Clipping Before the Limiter
Many professional mastering engineers use a soft clipper before the limiter to handle peaks more transparently than limiting alone.
A soft clipper rounds off the peaks of the signal rather than catching them with gain reduction.
This can allow you to reach higher average loudness levels without the limiter having to work as hard.
The result is often a master that sounds more open and natural at a given loudness level than one that relied entirely on limiting to get there.
The trade-off is that clipping adds distortion. Used lightly, this distortion is inaudible or even pleasant. Used heavily, it becomes obvious and harsh.
If you want to try this approach, add a clipper plugin before your limiter with the ceiling set 1 to 2 dB above the limiter threshold, and reduce the limiter’s input gain by the same amount.
Adjust by ear.
The clipper should be handling the brief, sharp peaks while the limiter handles everything else.
Metering Your Master
Use a loudness meter that shows LUFS integrated, short-term LUFS, true peak, and loudness range (LRA).
The integrated LUFS is your target level. The true peak shows whether your ceiling is correct. The loudness range (LRA) shows how dynamic the master is.
A very low LRA (below 4 LU) suggests over-limiting. A typical value for modern popular music is 6 to 10 LU.
Run the full track from start to finish before reading the integrated LUFS measurement.
Spot-checking only a section will give you a different number than the full track average.
A/B Your Master Against the Unprocessed Mix
The last step before calling a master finished is to compare it against the unprocessed mix at the same loudness level.
Use your makeup gain or output level to match the volume of both, then switch back and forth. If the master sounds notably better than the mix at the same level, the mastering chain is working.
If the master sounds the same or worse, something in the chain is either doing nothing or actively hurting the sound, and it needs to be reconsidered.
Related Resources