Smooth Doesn’t Always Mean Smoothed: The Hidden Setting Behind Every Frequency Response Graph
Here’s something that catches a lot of people off guard the first time they learn it: two completely honest, accurately measured frequency response graphs of the exact same product can look dramatically different, one looking wonderfully clean and gentle, the other looking jagged and full of small spikes, without either measurement actually being wrong. The hidden variable is something called smoothing, and understanding it changes how you should interpret basically every graph you’ll ever look at.
What smoothing actually does
Raw, unprocessed frequency response data, captured directly from a measurement microphone, tends to be genuinely messy, packed with countless tiny ripples, small notches, and minor irregularities across the frequency range. Some of these tiny wiggles reflect real, physically meaningful details about the product. Others are just measurement noise, background interference, or artifacts of the specific test setup that don’t actually represent anything the product is really doing. Smoothing is a mathematical process applied to that raw data, averaging nearby data points together to produce a cleaner-looking, more readable line, and it’s typically described using a fraction of an octave, like “1/3-octave smoothing” or “1/12-octave smoothing” (© hifireport.com — content licensed for editorial use).
The key thing to understand is what that fraction actually controls: the smaller the fraction, meaning the finer the resolution, like 1/24 or 1/48-octave, the less averaging is applied, and the more small details survive onto the visible graph. The larger the fraction, meaning the coarser the resolution, like 1/3 or 1/1-octave, the more averaging is applied, and the more small details get blended away into a gentler, cleaner-looking overall shape.
Why this matters so much for interpreting a graph
Heavy smoothing, like 1/3-octave, tends to look genuinely nice and tidy, and it’s easier to read at a glance, but it can meaningfully hide real, narrow resonance peaks, since a sharp spike concentrated in a narrow slice of the frequency range gets averaged together with the calmer areas immediately around it, shrinking or even completely erasing its visible presence on the final chart. Lighter smoothing, roughly 1/12 to 1/24-octave, preserves genuine narrow resonances much more faithfully, giving a more accurate, if visually busier, picture of what the product is actually doing (© hifireport.com — content licensed for editorial use).
This creates a genuinely important trap for anyone comparing measurements from different sources without paying attention to the smoothing setting used. A speaker measured and displayed with heavy 1/3-octave smoothing might look impressively flat and clean, while an identical speaker measured with lighter 1/12-octave smoothing might reveal a narrow, real resonance peak that the heavier smoothing had simply hidden from view. Neither chart is dishonest, exactly, but they can leave two very different impressions of the same underlying product.
Why manufacturers often lean toward heavier smoothing
It’s worth being candid about an uncomfortable pattern here: because heavier smoothing produces a visually flatter, more flattering-looking chart, it’s a well-known temptation, and sometimes an outright marketing tactic, to apply generous smoothing specifically to make a mediocre measurement look better than it really is. One widely referenced breakdown of headphone measurement practices explicitly demonstrated this effect, showing how the exact same raw earphone data, with real, significant imbalances in the bass, mids, and treble, could be made to look dramatically cleaner and more flattering simply by increasing the smoothing applied and adjusting the graph’s aspect ratio, without changing a single thing about the actual product (© hifireport.com — content licensed for editorial use). This is precisely why more careful, independent reviewers tend to favor lighter smoothing settings and are usually transparent and explicit about exactly which smoothing level they used, since that transparency lets readers trust the comparison is fair.
There’s also a genuinely legitimate reason for smoothing, though
It would be unfair to paint smoothing as purely a marketing trick, because it does serve a real, honest purpose too. Completely unsmoothed, raw data can include tiny irregularities that don’t actually correspond to anything a listener would perceive as meaningfully different, sometimes reflecting the limits of measurement equipment sensitivity rather than genuine product behavior. Human hearing itself doesn’t process sound with infinite frequency precision either; our ears naturally blend and average nearby frequencies together to some degree when perceiving loudness and tone, a real psychoacoustic phenomenon, which is part of why some measurement approaches specifically use variable smoothing schemes designed to roughly mirror how human hearing actually processes sound, applying different amounts of smoothing at different points across the frequency range rather than one flat setting throughout (© hifireport.com — content licensed for editorial use). Used thoughtfully and disclosed honestly, moderate smoothing genuinely can make a graph more representative of perceived sound, not less.
How to protect yourself as a reader
The most useful habit here is simple: whenever you’re comparing frequency response graphs from different sources, especially if you’re trying to seriously weigh one product against another, check whether the smoothing level is disclosed at all, and if it is, try to compare graphs using roughly similar smoothing settings rather than putting a heavily smoothed chart from one source directly up against a lightly smoothed chart from another. If a chart doesn’t mention its smoothing setting anywhere, that absence is itself worth treating as a small yellow flag, particularly if the curve looks suspiciously, almost impossibly clean and flat compared to typical measurements of similar products.
It’s also worth remembering that a genuinely well-designed product should still look reasonably good even under lighter, less forgiving smoothing settings. A speaker or headphone that only looks flat and clean when heavily smoothed, but reveals a mess of sharp peaks and dips once the smoothing is reduced, is telling you something real and useful about that product’s actual behavior, information that a heavily smoothed chart alone would have quietly hidden from view.
The takeaway
Smoothness on a frequency response graph is genuinely two different things wearing the same word: there’s the real, physical smoothness of a well-designed product’s actual sonic behavior, and there’s the mathematical smoothness applied afterward during the graphing process, which can flatter, obscure, or fairly represent that underlying reality depending on how it’s used. Learning to ask “how much smoothing was applied here” turns you from someone who can be easily impressed by a suspiciously perfect-looking chart into someone who can actually judge whether that chart is telling you the whole story.
Common Questions
**What is smoothing on a frequency response graph?**
Smoothing is a mathematical process that averages nearby data points together to create a cleaner, more readable line, typically described using a fraction of an octave, like 1/3-octave or 1/12-octave, with smaller fractions preserving more fine detail.
**Can smoothing hide real problems in a product’s measurement?**
Yes. Heavy smoothing, like 1/3-octave, can significantly shrink or even hide narrow resonance peaks and dips, making a product look cleaner and more flattering on a graph than its actual, more detailed behavior would suggest.
**Why do some reviewers prefer lighter smoothing settings?**
Lighter smoothing, roughly 1/12 to 1/24-octave, preserves genuine narrow resonances and irregularities more faithfully, giving a more accurate and transparent picture of a product’s real behavior, even though the resulting chart looks visually busier.
**Is heavy smoothing always a dishonest marketing trick?**
Not always. Moderate, disclosed smoothing can serve a legitimate purpose, since human hearing itself naturally blends nearby frequencies together to some degree, and some measurement approaches use smoothing specifically designed to better mirror how people actually perceive sound.
**How can I fairly compare frequency response graphs from different sources?**
Check whether each source discloses its smoothing setting, and try to compare charts using similar smoothing levels whenever possible, since comparing a heavily smoothed graph against a lightly smoothed one can create a misleading impression, even if both measurements are individually accurate.