We perceive tonal balance averaged over time. Returning to the earlier discussion of the painful HF that can occur when muting the LF, let’s now alternately mute the LF and then the HF.

If we do this slowly, the sound switches between bright and muddy, but if we do it more quickly, the “bright-dull” swaps begin to sound more - tonally - like a familiar kick-snare beat, which many of us rock music humans tend to find appealing.

The question is how rapidly we need to do the switching back and forth before our ears tell us it sounds O.K. Clearly this example has little realworld application, but what it highlights is that the averaging time we apply to real-time analysis systems is very relevant.

With music as a source, utilizing too slow of an averaging time can result in measured sound that looks “correct” but in reality, it alternates between sounding uncomfortably bright and overly murky. On the other hand, setting the averaging time too fast and short term can inspire us to EQ out frequencies that are not actually a tonal issue. My experience has been that an averaging time of 6 to 10 seconds provides a very useful visual readout that allows us to fine-tune the system EQ in real time while the artist is performing.

Comb filtering is not an issue with spaced sources. One of my pet peeves is that too much attention is paid to attempting to resolve comb filtering issues from separated sonic sources. Comb filtering occurs when two spaced sources are reproducing the same signal. When measuring a stereo system with a mono source signal, it’s all but impossible to get credible readings that parallel what our ears perceive.

Audio test gear will show a series of peaks and nulls, yet to our ears, it just sounds like sound coming from two places. This is not to say that comb filtering isn’t a major concern with sources in
close proximity or with lower frequencies, but on the other hand, if we test just one side of a stereo system at a time, we lose the ability to measure the combined coverage and see issues that are significant. So, what can we do to get our test gear to give us useful measurements of a stereo system?

Surprisingly there is a very simple (but rarely implemented) method. Rather than utilizing a mono random pink noise source, instead deploy two separate random pink noise sources. Even though the sources are tonally identical, because they’re independently random, they’re not the same signal. As a result, there will be not be comb filtering when running one pink source to the left side and the other to the right side.

For measuring low-frequency coverage, pan the two pink sources to the center, and for the most useful all-around measurements, I’ve found that panning the sources to about the 10 o’clock and 2 o’clock positions works well.

It’s my hope that these observations are helpful. But as with anything, check it, test it, refine it, and then bump it up a notch.

Dave Rat heads up Rat Sound Systems Inc., based in Southern California, and has also been a mix engineer for more than 25 years.

Page 2 of 2
Previous  < 1 2