CanJam NYC 2020: Audio Precision Measurements


To quote myself, “Measuring is an essential element in the design and manufacture of all audio components. But! As a tool for evaluation, or as a predictor of user satisfaction, today’s measuring procedures are almost useless.”

They are useless because audio components today are being measured while doing things that have very little relationship to what they were designed to do (i.e. play music).

That is why, every time Dan Foley of Audio Precision speaks at an audio show – I show up and listen. To me Foley is a kindred spirit in that he advocates for less focus on the frequency domain and THD+Noise; and for an expanded, more up to date, regimen of test procedures that would include multi-tones and real music (like dynamic drum tracks or piano chords).

Audio Precision interface example.

I asked Foley why people are not doing multi-tone testing and he said because manufacturers and their advertising departments prefer the tidy, easily-compared, (and low) numbers that THD+Noise deliver. He said multi-tone testing could only generate Total Distortion (TD) numbers; which would not directly correlate with existing THD+Noise figures. In fact, Foley said, components that delivered admirably low THD numbers might measure worse on TD than components with higher THD numbers. But! He said the multi-tone TD numbers have proven to correlate more directly with what listeners experience in their listening rooms.

When I got home I asked Stereophile’s technical editor why he didn’t do multi-tone amplifier testing and he said, “I am experimenting with tests like this, but the results with amplifiers are very hard to interpret.”

Foley also showed these extremely interesting “drum-track’ tests that were used to evaluate audio systems in very expensive cars.

For these regimens, the test signal consisted of a recorded sequence of drum-taps that were put through a car’s audio system and then recorded on a calibrated microphone in either the driver’s position, or a back seat – in the case of limousines. I love this test because you don’t get numbers, you get an image of the drum-track test signal going in superimposed over what the test microphone picked up. Every variation from the original input was obvious and easy to understand (it was mostly slow-starts and excessive ringing). This simple real-world test allowed audio-design engineers to make easy to evaluate adjustments – until the signal out looked most like the signal input. This test also showed how dynamic compression set in at increasing volume levels. I loved it.

I mentioned earlier how new headphone-only transducers like HEDD’s full-range AMT driver; which, along with RAAL’s SR1a full-range ribbon drivers, and the super-thin, super-light, super-stiff, full-range pure-beryllium drivers in the Final A8000 and DUNU Luna IEMs are leading the way to head-fidelity’s super low-distortion future. I look forward to hearing these sonic standouts more. See you next year.


Ortofan's picture

... square waves or measuring inter-modulation distortion a form of multi-tone testing?

What frequencies, relative amplitudes and waveform shapes is Dan Foley using?

TIM testing uses a combination of sine and square waves of differing frequencies.

Out-of-print magazines, such as Audio, High Fidelity and Stereo Review used to test speakers for the effects of dynamic compression, measure levels of disortion and show photos of the waveforms resulting from tone burst tests.

KaiS's picture

Measurements are done to visualize what and how much goes wrong in a transducer of any kind.

To choose the right type of measurement you need to know what you are looking for.
Every type of measurement has it's limits.
There are no shortcuts or easy solutions to cover everything with just one test.

A single sine wave e.g. can not show rattles and distortions that only occure on different frequencies.
The other extreme - a music signal as stimulus cannot show artifacts that are obscured by the stimulus itself.

Once acoustics and microphones are involved, things become really complicated, because the reception of a human's auditorial system is completely different to what a measurement microphone delivers.

Without being an expert in knowing what you do, measurements are pointless.
Laymen are easily fooled by nice looking numbers or curves that result from measurements.

This might be one reason why we don't see measurements of headphones here anymore, there is no expert present for doing it, and there is no common standard that can just be used.
It's not clearly defined what is right and what is wrong on a headphone.

Don't get me wrong:
I do not advocate not to do measurements, but the type and presentation needs to be well thought out to make them a valid guideline for everyone.

TomBreithaupt's picture

After 40 years playing with all this audio stuff I have one simply worded, but complex premise, summarizing all audio testing:

1) Products that sound good, almost always measure excellent.
2) Products that measure excellent have low guarantee they sound good.