Don’t fill the frequencies

Often I see advice given to fill up the frequency spectrum. That is, each instrument or part in the mix should fill a particular part of the frequency spectrum and that they shouldn’t overlap.

The problem with this approach is that most sounds – especially natural sounds – have a relatively wide frequency spectrum. For example, a human voice might have most of its energy in the upper mids (perhaps around 2.5kHz), but will also have useful information ranging from below 200Hz to above 20kHz. Similarly, pianos, drums, guitars and other instruments have wide spectra. Filtering this information out to keep only the most important frequency ranges will result in a sound that is artificial and unnatural.

On the other hand, this might be a useful approach for sounds that have no relation to natural acoustic sounds – such as synthesisers. Audio generated by frequency modulation synthesis or granular synthesis is nothing like any acoustic sounds, and so there is no expectation of what it should sound like.

Regardless of the instrumentation of your mix, it’s ok for sounds to overlap. Consider a mix with voice and piano. Both instruments share the same frequency ranges, yet we can hear both instruments clearly. This is because there is much more to sound than frequency. There are characteristics specific to the instruments, such as harmonic structure and envelope, and there are characteristics specific to the composition, such as harmony and rhythm. It it these characteristics that allow us to distinguish between different instruments – not frequency range alone.

Problems exist when two instruments share the same characteristics. For example, if a song has both an acoustic piano and electric piano, it may be difficult to distinguish between them. In this case, the most natural way of addressing this problem would be to remove one of the instruments (or at least only have one playing at a time). Simply filtering the sounds won’t change the characteristics enough, and doesn’t actually change any of the characteristics that are causing the problem.

A similar problem occurs when there are several vocal parts in a song. In this case though, the effect of making the individual parts difficult to distinguish is deliberate. All the vocal parts are designed to blend together. If there is one part that needs to stand out from the others (such as the lead vocal), this part is mixed in the foreground, and the other parts and mixed in the background. Rather than separating them by frequency, it is much more natural to separate them by depth.

-Kim.

Advertisements
  1. January 18th, 2010
Comments are closed.
%d bloggers like this: