What makes a cleaner mastering limiter?Posted on 13th October 2022
Boost is the mastering grade limiter from UrsaDSP. It is designed for transparent results in order to give the engineer more control over their sound.
To evaluate Boost’s audio performance, we measured the distortion it produces, and compared the results to three market-leading mastering limiter plugins.
Distortion and Clarity
Distortion, broadly-speaking, is a change to the shape of the audio waveform. Put another way, it’s the addition of harmonic content to a signal which was not there before.
In audio terms, distortion manifests as colouration or a loss of clarity. Whilst this can also add a welcome quality to sound, we believe that adding this should be a conscious choice of the producer / engineer and not mandated by the limiter.
We measure distortion by looking for added harmonics.
Types of Distortion
By using a single pure tone at 12dB over full scale, we can trigger each limiter in turn and see what distortion is introduced when reducing the level.
We analyse the output of each limiter to see how many extra harmonics are added and at what level. We also use their distribution and layout to identify the types of distortion present.
An ideal output would be the same signal 12dB quieter with no additional harmonics.
Here we have put the same track through Boost and the 3 other mastering limiters.
They are all targeting -12.0 LUFS with sufficient peak headroom to match typical streaming platform guidelines.
Making your own comparisons
You can test Boost against other limiters using a test tone generator and a frequency analyser. We’ve set up this handy Reaper project that will work with the Boost demo and provides a place for you to add your chosen limiter.
CPU usage and latency compared
In True-peak mode, Boost compares favourably with the other options – showing less latency and below average CPU consumption.
In Normal mode it is by far the most efficient.
Note that Other Limiter 1 is not a true peak limiter and Other Limiter 2 is shown in its (default) 8x oversampling mode as it does not have an explicit true peak mode.
True peak mode is only required for the last plugin on the mastering chain as otherwise any plugin with non-linearities is likely to introduce more intersample overs.