THD Distortion Audibility Test

Click Start Test to run the THD Distortion Audibility Test, a blind staircase that plays a clean sine tone against versions carrying progressively less harmonic distortion — pick whichever one sounds different using the First or Second buttons. The test narrows in on the exact THD threshold where you can no longer hear the difference, a number that says as much about your playback gear as it does about your ears. The check whether your vocal range is working properly gives you a clear answer instead of guessing.

Run a THD distortion audibility test and you'll quickly discover a fact that trips up most spec-sheet shoppers: a low total harmonic distortion number doesn't automatically mean your ears will register anything different. What actually decides whether it's audible is a threshold set by frequency, loudness, masking, and harmonic order — not the bare percentage on a datasheet. This guide breaks down how those thresholds get measured, why 0.001% THD can sound identical to 0.1% THD under the right conditions, and what real listening test results reveal about the limits of human perception.

What Is a THD Distortion Audibility Test?

A thd distortion audibility test starts with a simple measurement: total harmonic distortion (THD) is the ratio between the energy in a signal's unwanted harmonics and the energy in its fundamental tone. Feed a pure sine wave through an amplifier, loudspeaker, or DAC, and any nonlinear distortion in that device adds new components onto the signal's spectrum — those extra components are the harmonics. The standard formula looks like this: Try the free check whether your dolby atmos speaker is working properly for a quick, no-install way to check this yourself.

$$THD = \frac{\sqrt{V_2^2 + V_3^2 + V_4^2 + \cdots + V_n^2}}{V_1} \times 100\%$$

Here \(V_1\) is the amplitude of the fundamental and \(V_2\) through \(V_n\) are the amplitudes of each harmonic above it. A device that adds mostly high-order harmonics at low energy can measure with a tiny THD figure and still sound worse than one with a slightly higher THD dominated by low-order harmonics, because your ear doesn't weight every harmonic equally.

That's the entire reason audibility testing exists: harmonic content on a spec sheet doesn't tell you what a listener will actually notice. A related figure like SINAD folds residual noise into the same ratio, which is why a device can post an excellent single-number reading while its underlying harmonic structure is still audible once you run it through proper masking analysis — a single-number reading is never a complete picture of sound quality.

How Audibility Thresholds Are Measured

There's no single audibility threshold for distortion — it shifts with frequency, level, program material, and harmonic order, which is exactly why researchers plot detection curves by pitch instead of quoting one flat number. The audibility of distortion at a given pitch depends on how close the harmonic content sits to a much louder tone, and on how close that tone sits to a masking threshold shaped by everything else playing at the same time. Use the speaker phase alignment test whenever you want a fast, repeatable way to confirm this.

Auditory Masking and Masking Curves

Auditory masking is the reason two devices with very different measured distortion can sound the same. When a loud sound occupies roughly the same pitch region as a quieter one, the quiet sound simply disappears — a soft harp part sitting close to a loud trombone section gets buried entirely, no matter how clean the recording chain measures on a bench.

Masking curves plot exactly how much extra level a harmonic needs above the primary tone before it becomes detectable, and the foundational work behind this — Zwicker and Fastl's research in psychoacoustics — remains the basis most modern masking models are still built from. Because masking is strongest close to the frequency of the masking tone, mid-range distortion hides far more easily than distortion near the edges of what you can hear.

The Absolute Threshold of Hearing

Below a certain sound pressure level, a tone can't be heard regardless of masking — that floor is the absolute threshold of hearing (ATH), typically modeled on the ISO 226 equal-loudness curve. Your hearing threshold rises sharply at low pitches: a harmonic at 100 Hz needs a far higher dBSPL level to register than the same component at 1kHz, a frequency where sensitivity roughly peaks. This is measured against a calibrated reference — a full-scale reference tone at 0 dBFS mapped to a known dBSPL playback level — so a result expressed in decibels translates consistently between a bench measurement and an actual listening room.

• Pitch: low-frequency content hides more easily than high-frequency content, since your hearing threshold rises sharply below a few hundred hertz.
• Masking: a louder tone close to a harp or trombone in pitch can bury the quieter one completely, regardless of how clean the audio equipment measures on a bench.
• Noise floor and dynamic range: content sitting below a recording's background noise or a system's dynamic range stays inaudible no matter what a THD test reports.
• Playback level: loudspeakers, subwoofers, and headphones each reveal it differently depending on sound pressure level and listening loudness.

Pure Tone and Two-Tone Distortion Test Methods

Because SINAD and THD+N don't apply any perceptual weighting to individual harmonics, the only reliable way to define a threshold is to run a dedicated distortion test with controlled signals: pure tone and two-tone sequences. A pure tone — a single sine wave — works cleanly below about 1.5 kHz, but you can't test how audible distortion is near the top of the range with one such tone alone, since its harmonics would fall outside what you can hear. Above that point, testers switch to a paired-tone signal — two sine waves played together — and measure the result against an equivalent-amplitude reference tone so it stays comparable to a standard harmonic reading.

Just-noticeable thresholds vary sharply by test signal, harmonic order, and pitch, as this comparison shows:

A special worst case shows up at 1 kHz: combining 3rd order and 5th order components at equal level tightens the threshold further, while adding a 7th order component on top changes the picture only slightly below 1500 Hz.

Hard Clipping and Just-Noticeable Distortion Levels

Hard clipping is used as the reference worst case in most audibility research because it generates the densest harmonic content of any "normal" nonlinear distortion (polarity inversion and total signal drop-out are worse, but they aren't representative of typical gear). Odd harmonics dominate a hard-clipped signal, while even harmonics are largely absent from a symmetrically clipped waveform — the opposite is true of many naturally occurring nonlinearities, like certain tube-amplifier stages, which is part of why two circuits with the same measured THD can still sound different. Because clipping pushes so much energy into high-order components, it produces some of the lowest just-noticeable distortion levels of any nonlinearity type: smaller amounts of clipping-type harmonic content are audible compared to smoother, low-order content at the same measured percentage.

A few underlying factors shape where that threshold actually lands. Bandwidth matters: measurements are only valid over a specified range, and a moving filter that tracks a fixed multiple of the test frequency keeps results comparable across the full frequency range. Phase shift mostly doesn't matter — your ear is largely insensitive to it with complex signals — but a hidden polarity inversion can distort a signal indirectly, since cone displacement in a loudspeaker driver is rarely symmetrical to begin with. Researchers describe the resulting margin as a signal-to-distortion ratio rather than a flat percentage, because it's the ratio to the primary tone — not the harmonic content itself — that predicts whether a listener actually notices it.

SINAD, THD+N, and Why Lower Numbers Aren't Always Better

In the 1970s, manufacturers chasing ever-lower THD numbers on early solid-state amplifier designs pushed distortion components down by piling on negative feedback. One unintended result was Transient Intermodulation Distortion (TIM) — a fast, high-frequency artifact that only showed up on musical transients, and that never registered on a static single-tone THD reading at all. A device could hit a vanishingly small SINAD or THD+N figure on the bench and still sound worse in a listening room, because those single-number audio measurements say nothing about how the harmonic content is distributed across the spectrum.

If your own headphones already introduce 0.1% THD, a perfectly clean signal (0% THD) will still reach your ears carrying that 0.1% distortion. You can't perceive a difference between two sources that both sit below your own playback chain's distortion floor.

Detection Threshold and Distortion Detection in Practice

Reliable distortion detection therefore needs more than a single-pitch reading. A meaningful -6 dBFS reference-level test should sweep the full range rather than just the industry-standard 1 kHz point, and should be evaluated against a masking-based model rather than a flat percentage. Only then does a detection threshold reflect what a listener in front of real audio equipment will actually notice, rather than what looks best on a datasheet.

Real-World Listening Test Results: Loudspeakers, Subwoofers, and Music

Controlled listening sessions with actual music back up these bench-derived thresholds almost exactly. In one listening session, groups of listeners of varying ages were played rock and pop music through a pair of speakers and a subwoofer while pure tones were mixed in at gradually increasing levels, until each listener signaled that something didn't sound right. The tones ran from 20 Hz up to 10 kHz, at playback levels between 86 and 92 dB SPL — loud enough to be realistic, quiet enough to be comfortable.

The results were strikingly consistent with the masking predictions above: at low pitches, listeners were nearly deaf to the added tones. Content at 20 Hz and 40 Hz had to be raised louder than the music itself before anyone noticed it, while content around 280 Hz and below could reach roughly 20% before it became audible against the music.

Only at higher pitches — 5 kHz and up — did detection settle close to what a bench THD+N reading would suggest, around 3%. None of this changed with the specific selection of music, the playback order, or whether listeners knew in advance what to listen for, which strongly suggests the effect is perceptual rather than a quirk of one test setup — and it held whether the procedure ran as a formal blind test or not.

The takeaway for anyone comparing amplifiers or vinyl versus digital playback: the number on a spec sheet is a starting point for sound quality, not a verdict. Two components with wildly different THD figures can be indistinguishable in music playback, while a lower-reading unit dominated by the wrong harmonic order can occasionally sound worse. Even legacy formats with plenty of measurable noise — vinyl's groove noise and tape hiss included — remained listenable for decades precisely because so much of that noise sat below what real program material could reveal.

What a THD Test Result Actually Tells You

A THD reading is a starting filter, not a final verdict. It tells you how much unwanted energy a piece of gear adds under one specific, controlled condition — usually a single sine tone at a standard pitch — and nothing directly about whether that energy will survive masking by real music, sit above or below what you can actually hear at that pitch, or land in a harmonic order your ear happens to be sensitive to. Reading the result correctly means asking where the energy sits across the range, what order the dominant components are, and how far below the program material's own noise floor it falls — not just comparing two percentages and assuming the smaller one wins.