Why smooth freq curves?

Ive heard people mention that equaliser freqency curves should be smooth without drastic peaks or troughs. I would like to know why and if it is more important at the hi or lo freq range. I just dont get it.


Sounds in nature are made up of many different tones, overtones, and harmonics. This is an illustration of a single piano note.


It’s the note “G” plus the sound of the wood, size of the other strings, position of the pegs, etc.

This is why you can’t get rid of one single instrument in the orchestra by “filtering out” the single note they’re playing. You get the fundamental but not the harmonics.

If you gently equalize a sound, you can make it muffled or crisp without seriously changing the instrument. If you start equalizing in sharp spikes or edges you can change the relationship of the overtones and the quality of the sound. The difference in different makers of violins is all in the overtones. They will all play A#, but they won’t all do it with the same quality.

If you get really exotic, you can change the volume of the sound accidentally. Overtones and harmonics don’t just add up. Some of them naturally subtract and if you equalize those out, the volume of what’s left can increase causing overload and distortion.


Wow! Great answer. Ok i think i understand it. Subtle eq is preferable.

In Audacity’s eq editor, the length of filter(green line) smooths more at the low end. Is this because the overtones are further apart there?

But the manual states “…the smoother curve produced may actually sound better, unless you are modifying very low frequencies”.
Is this just referring to a sharp low cutoff?

I think im getting it now.

You don’t need to be obsessive about this. The crucial term is “drastic”. A little unevenness in the curve will have no audible effect. A “drastic” (large and sudden) change may cause unexpected and undesirable results, in particular it may cause ringing at frequencies close to the frequency of the “drastic” change. Whether or not that “ringing” is actually audible depends on the sound that you are working with and how drastic the change is.

The Equalization effect is a very sophisticated kind of filter, called an “FFT filter”. To slightly oversimplify how it works, the sound is split into hundreds or thousands of narrow frequency bands. The number of frequency bands is set by the “Length of Filter” setting. The “longer” the filter, the more frequency bands there are. The frequency bands are evenly spaced across the full range of frequencies (for 44100 Hz sample rate, that is 0 Hz to 22050 Hz).

However, human hearing works a bit differently. We hear pitch in terms of “proportion” rather than “equal intervals”. An “octave” represents a doubling of frequency, thus 400 Hz will sound “twice as high” (one octave above) a frequency of 200 Hz, and a frequency of 4000 Hz will sound “twice as high” (one octave above) a frequency of 2000 Hz. In these examples, we can see that if we split the entire frequency range (0 to 22050 Hz) into bands of 100 Hz, then between 200 Hz and 400 Hz there are just 2 bands, whereas between 2000 and 4000 Hz there are 20 frequency bands. Therefore, to achieve fine frequency control at low frequencies, we need the FFT filter to split the sound into many more (narrower) frequency bands.

And again, you can force that to work, but increasing the work also increases the time. If you want a particularly difficult effect on long show, you may have time to go out for coffee.