Can We Really Measure What We Hear?

The Limits of Microphones and the Power of Human Hearing

In the world of high-end audio, few debates stir more quiet tension than this one: if you can’t measure it, does it really exist? Or more pointedly: can the ear be trusted when the graph shows no difference?

Some engineers, even seasoned mastering professionals, continue to argue that everything we hear must be measurable using microphones and FFT-based analysis. It’s a comfortable idea.

Measurements are repeatable. Scientific. Free of bias.

But what if the most important things we hear aren’t easily measurable at all?

Our Ears Hear in Time, Not Frequency

Most measurement systems rely on frequency-domain analysis: spectral plots, waterfall graphs, THD curves, impulse response averages. These are useful, but they often ignore or smooth over the most critical parameter in music: timing.

Our brain doesn’t perceive the world through static frequency bins. It reconstructs reality by analyzing when things happen, on the scale of microseconds. This includes:

• The exact moment a transient starts
• The phase relationship between harmonics
• The timing gap between reflections
• The alignment of stereo image and reverb tail
• The rhythmic feel and internal coherence of instruments

As neuroscientist Brian C. J. Moore has written, “The temporal fine structure of sound is essential for localization, pitch perception, and timbre recognition.” (Moore & Sek, 2009)

And as MQA’s Bob Stuart and Peter Craven have shown, distortions in this timing, called temporal smearing, can result from digital filters, jitter, re-clocking errors, or simply poorly controlled analogue stages.

The Wisdom Behind the Quote

A famous saying in audio design goes: “If it measures good and sounds bad, it is bad. If it sounds good and measures bad, you’ve measured the wrong thing.” This quote is often attributed to Daniel R. von Recklinghausen, and it cuts through the heart of this debate. It doesn’t reject measurement, it challenges us to ask what we are measuring, and why.

If a change in a power supply, USB reclocker, or vibration footer clearly affects perceived spatial focus, transient realism or timing, but the graph shows no difference, then perhaps the graph isn’t looking in the right place.

As Bob Stuart notes in his SoundBoard article: “We are often measuring the wrong things, or not measuring what really matters to the brain.”

Research That Supports What Audiophiles Hear

There is now a growing body of serious research showing that our hearing is far more precise in the time domain than traditional measurements assume:

• Kunchur (2007) – Listeners can detect time misalignments as low as 6 microseconds
• Moore & Sek (2009) – Temporal fine structure is crucial for pitch and localization
• Unser & Aldroubi (1993) – Introduced B-spline filters to preserve timing, later adopted in MQA
• Stuart & Craven (AES 2019–2024) – Showed how downsampling and poor filter design lead to cumulative time-domain distortion
• Oohashi et al. (2000) – Ultrasonic content affects brain activity, even when not consciously heard

These findings reinforce what many listeners already know: just because a file or signal measures flat doesn’t mean it sounds natural or real.

Why Microphones Fall Short

A measurement microphone captures air pressure at a point. It doesn’t replicate how the human brain processes:

• Binaural time differences
• Directional filtering from the ear shape
• Expectation, memory, and emotional weighting
• Cross-modal perception (how vision affects hearing)
• Spatial modelling and source separation

Nor does it reflect how small distortions, even if buried under the noise floor, can affect coherence and intelligibility in real-world listening.

As MQA’s research shows, jitter artefacts as small as a few hundred picoseconds can shift the apparent position or timing of a transient. Most microphones won’t detect this. But your brain will.

Conclusion

Can we measure everything we hear?

Not yet. Maybe not ever, not in full, and not in a way that matches the brain’s elegant time-domain processing. But that doesn’t invalidate our perceptions. It challenges us to refine how we measure, and to accept that listening is not guessing, it’s perceiving.

As von Recklinghausen warned: if it sounds good and measures bad, you’re probably measuring the wrong thing.

Until the measurement instruments catch up, I’ll trust the one I was born with. 🎶👌😀

Peter Veth