October 2025

Not Even The (Audiophile) Earth Is Flat
The myth of 'Flat Frequency Response'.
Article By Roger Skoff

  Back well over half a century ago, that's 50 years, when I was a kid Hi-Fi Crazy, I had no money (kids never do), but I had been bitten badly by the High-Fidelity Audio (Hi-FI) bug, and – if I couldn't buy all of the toys and goodies that I so thoroughly lusted after, I could at least learn about them. So, like, I suspect, so many of us, I read everything I could find about Hi-Fi and Hi-Fi gear and started collecting (Yes, and even memorizing) "spec" sheets.

Most manufacturers don't provide them nowadays but, back then, everybody, and I and all my young Hi-Fi Crazy friends put together our own libraries of spec sheets and product literature,  and would sit around listening to our used, hand-me-down, DIY, or kluged together "systems" and learnedly discuss the wonders and glories of all the stuff we couldn't afford.

In the beginning (just to let you know how far back this was), everything was mono(phonic), but as time went on, it all – especially after 1957, when stereo LPs first came out –became stereo gear that we would dream of and pontificate about. Even so, other than going from a single channel to two, the specs of our dream systems pretty much remained the same:

The very first thing on the wish list was, of course, flat frequency response, and we all thought that if our system could just do 20Hz to 20kHz (We called it "CPS" – Cycles Per Second, back then) within plus or minus 3dB (decibels) from "flat", we would have sonic nirvana.

Distortion, of course, was another important criterion, but that was when tubes were the only choice, and 1% Total Harmonic Distortion (THD) was considered all that was either achievable or necessary. Sometimes an Intermodulation Distortion (IM) figure was thrown in, too, but it was usually considered secondary.

Do you know how Enjoy the Music.com employs those numbers that were chosen to be the industry standard and the stuff of audiophile dreams? It's easy: First off, the nominal range of ordinary human hearing is commonly stated as running from a lower limit (deepest bass) of 20 Hz (sinusoidal pressure waves in the air or other medium changing polarity from positive to negative and back 20 times a second) to (highest treble) 20 kHz (yup, the same pressure waves, but changing polarity 1000 times faster). That's what was used as the ideal response for any kind of audio.

Now, consider this: If you were to make a graph of the ability of a piece of gear to either produce (if it's a speaker), or to provide driving signal for a speaker (if it's part of the system's electronics), and were to mark the level of output for the same level of input at every frequency within the range of human hearing, if all of the output levels were identical, you would find that what you had created as your graph was a straight (flat) line. That's how a perfect piece of equipment would chart, but while being "flat" from 20Hz to 20kHz may be easy for a good piece of modern electronics, it's virtually impossible for an electromechanical device like a speaker or a phono cartridge. So, if "flat" is at least highly unlikely, what's good enough?

Do you remember earlier, when we were talking about a system that was flat within plus or minus 3dB across the entire spectrum of human hearing as being ideal? Let's go back and look at that "3dB" figure and see what it means.

Actually, it has two significant meanings: First, it represents either a doubling (twice as much) or a halving (half as much) of the actual sound level (the amplitude of the pressure waves in the air) as mechanical force. Second, because human hearing is not linear in operation, but logarithmic, it isn't perceived as being twice as loud, but is, in fact, just the minimum change of level, louder or softer, that's easily perceivable by the average person. (There are people out there who can hear a JND [Just Noticeable Difference] at as little as 1dB, but they're extremely rare).

Now we know where the 3dB spec came from, but what does it mean? Very little. Here's why: All saying that something will do "20Hz to 20kHz +/- 3dB" tells you is that the plot of the amplitudes of all of the frequencies in the stated range won't vary A total of more than 3dB above or below flat at any point on the graph. But it can all be diddled by where you choose to draw the "zero" line. So, if the total range of amplitude variation within the frequency range is 6dB and I draw the zero line so that it's exactly halfway between the highest level and the lowest level, my graph will show performance flat within 3dB across the range.

But if the bass, for example,  is 6dB down from the highest positive point and I draw the zero line through the highest point, my graph – even though the bass is still down 6dB – will show the whole range to be flat within 3dB. Same thing with every other frequency: By skillfully choosing where to place the zero line, you can have a whole series of peaks and valleys in the frequency response – have clearly audible 6dB differences all over the place – and still be able, a long as the highest level is no greater than 6dB louder than the lowest level, to truthfully claim frequency response "flat within 3dB".

Here's another way frequency response charts can be diddled – this one in noticed by Enjoy the Music.com's editorial team and quite often used by phono cartridge manufacturers to make their products look perfectly flat, regardless of their actual performance: Just print the graph on a single sheet of paper, perhaps two inches high by 18 inches long, with the logarithmic display of frequencies along the length of it and the amplitude levels printed on the vertical side, in 10 or 20 dB increments. Done that way, even a moving coil cartridge with a colossal resonant peak in the 35kHz range will still come out looking razor flat, and customers will always be impressed.

Impressive or not, real or not, even a theoretically perfect system, with dead-flat, distortion-free electronics, will still end up being played through speakers in an acoustic environment that may not be perfect. And even if the speakers, themselves,  are brilliant designs that tested perfectly flat in an anechoic chamber, that makes far less difference than you might imagine: You're not going to be playing them in an anechoic chamber, but in your own listening room, at home or elsewhere. And that room is going to have its own characteristics that will affect the sound of the speakers and the entire system, creating something other than what any outside testing might indicate.

This is not anything other than the absolute truth, and what it means is that, as with everything else about your system, don't worry about the specifications. Instead, concern yourself with what things sound like – preferably in your own room, listening through your own system.

Not only is perfection not generally possible, it may not even be desirable: Music played in an anechoic chamber sounds absolutely awful.

Listen. Use your ears. Pleasing them is what you're spending all that time, money, and effort on.

The only thing that's really important is if you...

Enjoy the music!

 

Roger Skoff