Take it on faith: A cease-and-desist letter to those who only believe in measurements

When I refer to measurements, I’m talking about acoustic room behavior, not gear specs — early reflections, decay times, and frequency response. Once speakers are in a room, those factors dominate what we hear.

Where I agree with much of the criticism is that current tools like REW and OmniMic often present measurements as isolated charts. Each chart can be useful on its own, but for newcomers especially, they create visual overload, lack perceptual context, bury left/right symmetry issues, provide no clear “good enough” stopping point, and rarely point toward concrete corrective actions. The result is a collection of data islands with no causeways connecting them.

My approach is an attempt to build those connections.

I start by measuring the three main acoustic pillars — early reflections, decay time, and frequency response — using consistent frequency banding. For each pillar, I evaluate left/right symmetry and left/right deviation from a target. Those deviations are then converted into severity scores that reflect perceptual audibility, not raw magnitude.

This matters because audibility is not uniform. A small mismatch at 2 kHz can be far more obvious than a much larger difference at 40 Hz. Likewise, a loud, asymmetric early reflection in the first few milliseconds is far more damaging to imaging than a later, quieter one. Perceptual weighting accounts for this by emphasizing deviations where our hearing is most sensitive and down-weighting those where it is more forgiving.

Because these three pillars are normally expressed in incompatible units — milliseconds, decibels, and energy-versus-time plots — their deviations are first normalized to a common scale (for example, 0 = within tolerance, 1 = clearly problematic). Once normalized, a decay issue, a reflection asymmetry, and a frequency-response deviation can be compared directly — not by how dramatic they look on a chart, but by how likely they are to be heard.

That normalization makes prioritization possible. Instead of asking “What looks bad?”, the question becomes: What is most likely audible, and what should I address first?

Once issues are ranked by perceptual severity, measurements stop being abstract graphs and start functioning as a map. High early-reflection severity points toward sidewalls or front/back walls. Overlong bass decay points toward corners and boundaries. Frequency-response issues, once major asymmetries are addressed physically, often point toward placement refinement or DSP/EQ.

Used this way, measurements don’t replace listening — they give structure to it. They connect the charts, reduce guesswork, and help translate data into decisions rather than endless tweaking.

That’s the intent behind the weighting, normalization, and prioritization — not faith in numbers, but a way to make them more useful.

Great post!

Exactly what i lived through when i tuned my room using mechanical resonators, tuned by ears and located at specific place in bundles...

Instead of using varieties of DSP i used my ears, in the process i learned basic acoustics with my "body" and ears ...

Acoustics rule audio experience not so much  the specs of the amplifier which matter for sure for synergy reason with the system...

When I refer to measurements, I’m talking about acoustic room behavior, not gear specs — early reflections, decay times, and frequency response. Once speakers are in a room, those factors dominate what we hear.

Where I agree with much of the criticism is that current tools like REW and OmniMic often present measurements as isolated charts. Each chart can be useful on its own, but for newcomers especially, they create visual overload, lack perceptual context, bury left/right symmetry issues, provide no clear “good enough” stopping point, and rarely point toward concrete corrective actions. The result is a collection of data islands with no causeways connecting them.

My approach is an attempt to build those connections.

I start by measuring the three main acoustic pillars — early reflections, decay time, and frequency response — using consistent frequency banding. For each pillar, I evaluate left/right symmetry and left/right deviation from a target. Those deviations are then converted into severity scores that reflect perceptual audibility, not raw magnitude.

This matters because audibility is not uniform. A small mismatch at 2 kHz can be far more obvious than a much larger difference at 40 Hz. Likewise, a loud, asymmetric early reflection in the first few milliseconds is far more damaging to imaging than a later, quieter one. Perceptual weighting accounts for this by emphasizing deviations where our hearing is most sensitive and down-weighting those where it is more forgiving.

Because these three pillars are normally expressed in incompatible units — milliseconds, decibels, and energy-versus-time plots — their deviations are first normalized to a common scale (for example, 0 = within tolerance, 1 = clearly problematic). Once normalized, a decay issue, a reflection asymmetry, and a frequency-response deviation can be compared directly — not by how dramatic they look on a chart, but by how likely they are to be heard.

That normalization makes prioritization possible. Instead of asking “What looks bad?”, the question becomes: What is most likely audible, and what should I address first?

Once issues are ranked by perceptual severity, measurements stop being abstract graphs and start functioning as a map. High early-reflection severity points toward sidewalls or front/back walls. Overlong bass decay points toward corners and boundaries. Frequency-response issues, once major asymmetries are addressed physically, often point toward placement refinement or DSP/EQ.

Used this way, measurements don’t replace listening — they give structure to it. They connect the charts, reduce guesswork, and help translate data into decisions rather than endless tweaking.

That’s the intent behind the weighting, normalization, and prioritization — not faith in numbers, but a way to make them more useful.

@kevinzoe I agree with @mahgister, that was a very good post.

kn

It's been stated here in different ways, but I know measurements are often used to argue one speaker is good and another is bad. This argument fails to measure things such as the recordings or my ears. So when people pull the 'science' card, they can in fact be completely wrong because they failed to account for everything.

@bartsw 

So when people pull the 'science' card, they can in fact be completely wrong because they failed to account for everything.

Also whenever anyone uses the phrase "the laws of Physics", it reveals a naive understanding of the subject. Physics has long moved on from absolute laws laid down by God to models of reality with inherent limitations.