Comparative Techniques In Audio Mastering
As a Mastering engineer, two of the most important considerations when working with artist music are:
Have we improved the aesthetic of the sound overall?
Have we taken away or masked any positive aspect in the essence of the original mix?
To effectively understand these considerations, it is important to remember small adjustments in the mastering scenario will have a large overall effect on the processed audio. For example, if we apply a 1dB boost to our session mix at a given frequency and filter type, it is in principle equivalent to applying the same filter boost to all the individual tracks in your mix. This could be 24 or more channels - a lot of EQ! Hence we must be considered in how we assess our processing to truly hear the change in the aesthetic action.
The following series of three comparative techniques can help audio engineers to answer these questions truthfully and accurately by interrogating the audio, and importantly the processing applied.
Remember we’re mastering, not mixing. We’re making a session mix that already sounded as the artist or producer intended ‘fit for purpose’, correct for its output format; enabling the music to translate effectively to all environments. Inherently this means we will make it sound balanced overall in its tone, spatial impression, and dynamic – Mastered.
Equal loudness:
The principles of how the human ear reacts to loudness laid out in the work of Fletcher-Munson dictate if we make something louder it will sound fuller; perceptively it has more weight or energy. Basically, if we turn it up, it will sound better! This can be easily proved by placing the same audio on two different tracks and simply turning one up by a couple of dB. Solo between the two, and the louder will sound better because our ears are fooling us! Of course, we know it is the same audio, it is just our perception of the playback that has changed.
For years, audio experts have been aware of this phenomenon. Transfer/mastering engineers have been striving to make their output format sound louder since the invention of the first compressors, first to enhance signal-to-noise ratio and minimize distortion, but also because it will sound better in comparison to another song. On jukeboxes and some radio stations in the early days of the record industry, this was the situation. The lure to record companies is that if your album was mastered louder, it would sound better in comparison to playback. This has resulted in what has been dubbed "The Loudness Wars" in recent history, but that's a topic for another day.
Equally, engineers in broadcasts initially used limiters to maximize loudness to boost transmission distance and quality by overload reduction, but they have always tried to make their program output louder than another’s – as it will be more catching as it perceptively sounds better. When flicking between stations on the radio, the more active or louder stations become more attractive to the ear. Louder adverts on TV make us more aware of their presence and so on. Even today with the advent of broadcasting standards engineers play with the sound balance to achieve this competitive loudness.
Even some manufactures have used this effect to enhance their equipment’s appeal on the initial demonstration by making the output of a processor slightly louder when all the controls are set flat – hence when turned on and bypassed in / out, it makes the sound perceptively better even though it may not be the case, it is just louder!
In good audio practice when applying to process, especially in mastering, we need to compare back to the original audio at an equal loudness to hear how the audio has actually been changed. If we boost an EQ we’ve made it louder, hence, almost always irrelevant of the actual change in tone, at smaller values (3db and below, 3db being a lot of EQ in a mastering scenario) our ears on comparison without balancing the audio (i.e. turning the EQ boosted audio down), the EQ’d audio will sound better. Equally the opposite would apply if we cut the EQ without boosting the process track the original will sound fuller etc.
Hence where adjustments are being applied, it is important to compensate for the boost or cut with EQ or use the make-up gain with dynamics processes to balance the output level. Facilitating ‘equal loudness’ in comparison to the processed / unprocessed audio - meaning we can hear the true adjustment in the processed sound. This technique of ‘Equal Loudness’ should also be applied when comparing the final mastered file back to the original. It is vital to make sure we have answered our main mastering considerations:
Have we improved the aesthetic of the sound overall?
Have we taken away or masked any positive aspect in the essence of the original mix?
Only then, in that true comparative can we honestly assess if we have made the artist’s music sound better in all regards. Equal loudness should also be used in some parts of the mixing process where we are making more subtle adjustments to a track. If we apply a lot of processing because we are being creative, let’s say we’re making a signature sound, using a pass filter, or just being very corrective, it is less important as we can hear how we have changed the sound creatively without the need for comparison at equal loudness.
Sum/Difference:
This is an invaluable comparative tool to facilitate understanding how the ‘tool’ applied has changed the original audio as we can hear what has been added or subtracted from the overall audio processing. You can create a Sum / Difference Matrix in any DAW with full delay compensation by creating another track below the original audio and assigning this to the same output. Directly copy the audio down to the new track, this has to be sample-accurate (use snap if necessary). In some DAW’s you can do all this by simply duplicating the original track.
Polarity reverse the new track, sometimes called phase invert – phase actually means time differential, or you could say delay, but polarity reverse is the correct term to use in this case. If you don’t have a polarity reverse button on your DAW’s mixer, use a plug-in or process the audio offline. If both tracks are un-muted and you’ve followed the instructions above, you’ll hear nothing when the audio is played back. This is because the two audio tracks are in perfect opposite polarity – hence one cancels out the other. As one waveform goes positive the other goes equally negative.
If you insert a plugin on the process track (top) and restart playback, you’ll still hear nothing if the plugin is totally passive. If the plugin is affecting the audio in any way, you will hear what is added or subtracted from the audio. Remember you’re listening to the difference, you don’t know if it’s a boost or cut without further investigation.
Some modeled plugins, if replicating the original unit in detail, will evidence this difference (Pultec or Manley emulations are generally a good example). With an EQ plugin, if we now apply either positive or negative gain we’ll hear what is being added or taken away by the EQ, and the phase of the actual plug-in making the process. You can use this technique to hear how any process or combinations of processes have changed the sound. This technique is an excellent way to start to understand how a process works i.e. hearing the shape of an EQ, how its 'Q' width changes with increased/decreased gain, and so on. How dynamic is shaping the sound? Some plugins have this sum/difference built-in with a ‘solo’ function. The Weiss EQ1 is a good example, also most noise reductions have this function, where it can be often useful to fully perceive what has been removed to decide on how much reduction to apply.
If you’re using external analogue hardware in the chain you can ‘ping’ the external effects loop in some DAW’s to achieve the correct delay compensation or just record the process back in and manually line up. With ping, you can achieve the sum/diff in real-time. In practice now having a processed track on top, and an original audio track below (polarity reversed) we can, by muting just the bottom track, listen and conduct processing of the source audio. Un-muting the bottom track, we will hear the difference – the effect of the processing applied. By soloing the bottom track we’ll hear the original audio in comparison (assuming your audio program is set to defeat mute on a soloed track).
Remember – louder sounds better to our ears - you must make equal loudness comparisons to hear the true change in the audio when comparing between the original and processed. If you’ve readjusted the level, to make an equal-loudness comparison to hear the Sum/Difference you must adjust the audio tracks to the same relative output level by ear. Often, we can find this point especially with dynamics processing by ‘tapping’ or ‘nudging’ the level of one track up and down to find the point with the most ‘null’. Even 0.1db can make quite a difference to the amount of null. I use this technique every time I’m mastering a track or album – it makes me aware of how the processing is affecting the overall sound. Is there anything being removed or enhanced I didn't what to, and most importantly the question we should always ask ourselves – have we actually made the audio sound better? If we have not improved the sound overall then we have not done our job.
Mid/Side exploration:
Mid Side [M/S] is a powerful corrective tool in the Mastering engineer’s arsenal when it comes to correcting mix errors. Equally, when used as a comparative tool we can fully interrogate the stereo/mono components of a mix. When listening to a track for the first time, I’ll always listen to the (M) Mid (L+R – 6db) and the (S) Side (L-R – 6db). A good Mastering monitor controller will do this for you, or you can use a plug-in tool such as Voxengo MSED or BXsolo (both freeware). By listening to the ‘S’ in isolation we can establish the stereo elements of a mix and what can be enhanced to improve the overall image. Often less well-mixed tracks will lack width or have ambiance issues that can be greatly improved by the judicious use of EQ, and pre-emphasis at the Mastering stage.
The use of a Mid Side listening tool will enhance your ability to understand the session mix panorama, its elements, and help you to improve the sound imaging and clarity overall.
In Conclusion:
We need to make sure we’re not being fooled by our ears. They’re our best and should be our most trusted tool, though they can be our worst enemy without the application of good audio practice.
Every time I master an album, I’ll use all these comparative tools to help my abilities to creatively enhance the audio. If you start to integrate this good practice into your daily engineering, you will not only improve your understanding of sound as an engineer but start to produce more consistent, better-sounding mixes and masters. Though always remember - sometimes no process can be the best process. The use of comparative tools in practice will help you to truly answer our main aims of being a mastering engineer:
Have we improved the aesthetic of the sound overall?
Have we taken away or masked any positive aspect in the essence of the original mix?
