RealTraps - Audibility of Narrow-Band EQ

Dispelling a common audio myth


More RealTraps Articles




"The correlation between frequencies changed with EQ and the frequencies present in the music is very important!"











"Nulls of 20 to 30 dB or even deeper are common, and you'll blow up your speakers trying to raise such deep nulls enough to be flat."

By Ethan Winer

Many people believe that very narrow peaks and nulls are not audible, especially nulls. While a narrow bandwidth affects a smaller range of frequencies and thus less overall energy than a wide bandwidth, very narrow bandwidths can still be highly audible. What matters most is if the frequencies being boosted or cut align with frequencies present in the source material.

To illustrate clearly what is meant by bandwidth, the two screen shots below show an equalizer set to cut 165 Hz by 10 dB. The top image has a Q of 24 and the lower image has a Q of 2. In this case "Q" stands for a filter's Quality Factor, and a higher Q means the filter bandwidth is narrower.

I believe the misunderstanding arises from THIS 1981 paper in the AES Journal by Roland Bucklein where he describes testing the audibility of boosts and cuts at various bandwidths. Some of the tests used speech and white noise, but others used music. With white noise, the total change in energy affects our ability to hear the EQ. So a wide bandwidth boost increases more energy and is more audible than a narrow boost. Likewise for wide cuts that reduce more total energy.

But for the music tests, the frequencies boosted and cut in Mr. Bucklein's experiments did not align with the frequencies in the music being played. Music consists mostly of single tones (and harmonics which are also single tones), so the correlation between the frequencies changed with EQ, and the frequencies present in the music, is very important!

To illustrate the potential audibility of very narrow boosts and cuts I created a series of three short Wave files, each about 1 MB in size. The first is a clip from one of my pop tunes as I mixed it. The second is the same clip with 10 dB of very narrow EQ boost (Q = 24) at 165 Hz. The third is the original clip but with a 10 dB very narrow cut (Q = 24) at 165 Hz. I chose 165 Hz because that's an E note, which is the key of the tune. So in these examples, the boost and cut are very obvious because they align with notes the bass plays.

Original Clip
Clip with 10 dB boost
Clip with 10 dB cut

This is directly related to trying to use EQ to improve room acoustics at low frequencies. One big problem with using EQ to improve acoustic problems is it's not possible to counter deep nulls. Nulls of 20 to 30 dB or even deeper are common, and you'll blow up your speakers trying to raise deep nulls enough to get a flat response. I've seen EQ proponents claim that nulls are not a problem because they're so narrow, and they often cite the same Bucklein article!

However, it's true that the frequency response in a room can change drastically over very small distances, even at very low frequencies. So a deep null at one ear may be less deep at the other ear. But not all nulls are highly localized.

Hopefully the example files show clearly that even very narrow nulls can be damaging if they align with notes in the music. Top

Entire contents of this site Copyright 2004- by RealTraps, LLC. All rights reserved.
No portion of this site may be reproduced in any form without the express permission of the copyright holder.