Listening With Your Eyes and Mastering By Numbers?

A before (red) and after (black) comparison of a well-mastered track, played back at the same loudness level.

The only way to determine whether or not a track sounds good is to listen to it. Measuring it won’t cut it, nor will staring at a waveform on a screen.

This may seem like common sense, and we’re intuitively well aware of the dangers of the all-too-common trap of “mixing with your eyes”.

However, thanks to newly accessible audio technology and some fallout from the so-called “loudness wars“, an ever-increasing number of musicians, engineers, industry commentators, and even fans, are relying on the use of waveform comparisons to support their assertions that the track they’re listening to sounds “good” or “bad”.

What we should really be concerned with of course, is the sound, the whole sound, and nothing but the sound.

What’s in a Waveform?

One of the most common areas in which you’re likely to encounter an image of waveform in support of an assertion that a track sounds “good” or “bad”, is in the increasingly common debates over loudness in contemporary music.

While there are sound arguments to be made that the quest for excessive loudness can lead to a compromise in sound quality, it’s also possible to make an excessive and poorly-supported case against greater loudness when it comes to any given track.

This kind of flawed argument often starts with a screenshot of an older and more dynamic track, say, something from Jeff Buckley’s Grace. There it is: A visual waveform of “Last Goodbye“, shown in all its “peaky” glory, highlighting what a good-sounding, dynamic track should look like.

Then, the waveform of a track from deep in the trenches of the loudness wars will appear, let’s say Black Eyed Peas’ “Boom Boom Pow” from their album The E.N.D. This is what a “bad-sounding” track looks like. It is presented as visual evidence that loudness is a bad thing.

When you compare the two waveforms, you will see some similarities, and some obvious differences, too. What the two waveforms will usually have in common is their maximum peak level: “Last Goodbye” peaks at 0dBFS, as does “Boom Boom Pow”.

You can also tell, just by looking, that “Last Goodbye” has more pronounced peaks than “Boom Boom Pow”, and so we can confidently say that its “peak-to-average ratio” is far greater. “Boom Boom Pow”, for its part, appears to have peaks that aren’t much higher than the average level at all.

But what does this mean when it comes to sound quality, as perceived by the listener?

From this visual comparison—and from the knowledge that one track has a greater peak-to-average ratio—we can make the following assumption about the sound of Track A compared to Track B: Nothing.

Nothing at all can be concluded about the subjective quality of how a track sounds from how it looks or how it measures. That’s for your ears to decide.

Less Than Meets the Eye

Perhaps we could state that Track A has more dynamics than Track B, but so what? This has no fixed relation to the perceived sound quality of the track.

All else being equal, many productions may indeed sound subjectively “better” to many listeners when less dynamic range is present. That’s just a fact about human perception and differing tastes.

Countless terrible-sounding and incredibly dynamic tracks have been produced throughout history. Similarly, there are plenty of great-sounding tracks with relatively minimal dynamics.

Certain styles of music may even strike us as sounding “better” when produced to sound loud and compressed, and so they should be. There’s nothing wrong with loud tracks. There, I said it: Loud, bad-sounding tracks are a problem, yes, but so are quiet, bad-sounding tracks.

If you can make a good-sounding and loud record, then you should be praised for it. This is one reason why world-class mastering engineers are few and far between. And to be fair, for anyone but the most experienced and knowledgeable audio engineers, the louder they make a track beyond a certain point, the worse it is likely to sound.

Examples of Bad Mastering Do Not Mean Good Mastering Is Impossible

Not satisfied with mere visual waveform comparisons, some brave folks have even taken it upon themselves to try their hand at mastering to educate us as to what “Smells Like Teen Spirit” might sound like if it were released today and put up on YouTube.

They will have “re-mastered” the track to be more in-line with current tastes in level. Of course, they didn’t go back to the original 2-track analogue tape masters to actually re-master the track from the ground up, as is the standard practice in professional circles. They simply re-mastered an already mastered track, and without much of an effort (or the requisite experience) to make it sounds as good as possible while doing so. Predictably, it sounds absolutely terrible.

A relatively inexperienced YouTuber “mastering” an already mastered track by simply throwing a limiter over it is not evidence that loudness is our ultimate enemy, and that records being made today necessarily sound awful. This attempt at an argument is even more distorted and misleading than their attempt at mastering.

I’m reasonably sure that many of the people making these kinds of videos have never had a track professionally mastered or, if they have, never paid attention to what was really going on. There are certainly many cases in which a mastered tracks can sound much louder and much better. Here, only taste and experience can guide us.

As an example, Figure 1 shows an unmastered waveform of a portion of the track “Down at Your Buryin'” by George Faber. Figure 2 shows the same track, as mastered by Bob Katz, who is a staunch and vocal opponent of the pursuit of excessive and unnecessary loudness in music. A comparison of the difference in peaks is shown in Figure 3.

Figure 1: Unmastered Version

Figure 2: Mastered Version

Figure 3: Waveform Comparison

Figure 1 has many more peaks and a lower RMS level than Figure 2, and consequently has a higher peak-to-average ratio as well.

The pre-mastered version is far more dynamic. During mastering, the track has been compressed and its peaks have been limited. Although this mastered track is significantly less dynamic, it sounds far better to my ears, and I am guessing that most readers will agree.

If you’d like to hear these particular examples for yourself, in just seconds, you can sign up for a free account on Bob Katz’ site, Digital Domain, and then download the before and tracks from this master of mastering engineers by clicking here:

Once you’ve taken about 15 seconds to register for a free account at Bob Katz’ site, you can download these files to hear before-and-after clips from an undisputed “master” at mastering. Why the extra step? We respect copyright, and we respect Bob enough not to ask him to endorse a story that may not represent his own views on the matter.

The difference in visual peaks isn’t only the result of compression and limiting, of course. Other signal processing, such as equalization and mid-side adjustments will also have an effect. But based on Katz’ mastering style, it’s fairly safe to assume that there has been no added distortion from slamming the track into a limiter or intentionally clipping the A/D.

This kind of tasteful decrease in dynamics, which leads to increased loudness and an increase in perceived quality, occurs every single day in professional mastering studios. It’s what we strive to do. Some very successful mastering engineers push things much farther than Katz would be inclined to, but they are also able to achieve results that their clients, and their fans’, love.

It is also true that every day, a mastering engineer somewhere in the world will make a record sound both louder and worse. And there are indeed mastered tracks that sound better than the original version that aren’t any louder at all. But the bottom line is that every day in professional mastering houses, “tracks with lots of peaks” are turned into “tracks with fewer peaks”, and I would like to think that the vast majority of them end up sounding even better than before.

You simply cannot draw conclusions about the sound of a track, or about whether the mastering process has helped or hindered the sound by looking at a waveform alone.

If you stuck at the level of reasoning of the average YouTube critique however—“less pronounced peaks and higher RMS level equals worse sound”—you could be led to conclude, just by looking, that Figure 2 up aove is another casualty of the loudness wars and that it clearly must sound worse. This of course, is not even close to accurate. Not for this track, and not for so many mastered tracks that are significantly louder than it is.

In the end, only your ears, not your eyes or your conscious faculties of reasoning, can determine what sounds good to your ears. Here’s to closing your eyes and really listening.

Mark Bassett is an educator, and engineer, and a PhD candidate in Psychoacoustics at the University of Sydney.

Please note: When you buy products through links on this page, we may earn an affiliate commission.