CanJam New York 2019

Besides major new-product introductions, must-do auditioning, and big-time meet-n-greet, CanJam New York 2019 appears to have a theme, which just so happens to be one of my favorite topics: headphone measurements.

New York CanJam is always extra-exciting because not only does it put visitors in the center of today’s headphone universe; it puts visitors in the epi-center of the New York City experience. Think walking distance to Times Square, Rockefeller Center, and Central Park.

Doors open at 10:00 am, Saturday, February 16th. The measurement mindset kicks in three hours later at 1:00pm when CanJam founder Jude Mansilla ( presents a seminar entitled “Measuring Headphones Today: Precision and Challenges” in the Majestic Music Box room on the 6th-floor of the Marriot Marquis. Jude will examine the level of precision needed to accurately measure (and describe) the performance of today’s sophisticated headphones and IEMs. Modeling perception. How measurement data is obtained. Why headphone measurements matter and why they sometimes don’t. Jude will also explain the challenges today’s headphones present for those who want to measure them.

Jude’s seminar with be followed by Dan Foley from Audio Precision/Alma International presenting what I suspect will be a mind-opening talk entitled: “It’s 2019! So Why Are You Still Measuring Audio devices Like It’s 1969.” Ah dah! Good question Right? Dan Foley will explain how audio measurement techniques and choices concerning what to measure have barely changed at all in five decades. Focusing on how frequency response is still measured using a sine-wave stimulus, as is harmonic distortion. This presentation will introduce new measuring technologies, using SPEECH and MUSIC, for characterizing the performance of any kind of audio device. I have waited 100-years, and would walk 50-miles in Arctic cold, to hear this lecture. (You can be sure, I will be the man in black in the second row.)

At 4:00pm Rob Watts, of Chord Electronics, will continue in a similar vein, with a talk entitled, “Chord Hugo TT2 Technical Seminar” wherein he describes the thinking and technologies behind the new Hugo TT2 as well as a detailed discussion of its measured performance.

Rob Watts will continue his explaining on Sunday, at 12:30pm with a technical talk about the Chord Hugo M Scaler and the importance of transients for auditory perception and how digital audio has problems recreating transients accurately.

The second (and last) seminar on Sunday is titled, “Personal Audio AMA (Ask Me Anything) questions like: How do headphone designers achieve their desired tuning? Will Digital Audio ever approach analogue? And, last but most importantly (to me): Do headphone measurements really matter?

In between smart seminars, you can audition important product introductions like Qobuz streaming (free trial coupons will be given out at registration desk), Focal’s new and beautiful, beryllium-domed, closed-back, Stellia, the new JPS Labs Dianna Phi, Audeze’s amazing $300 Mobius, HiFiMan’s Ananda BT, and last but equally major, the radically-evolved Schiit Ragnorock 2, which, are you ready(?), comes with a remote. But that is nothing. Schiit will also be introducing their long-awaited Soul Turntable… And very much more at CanJam New York City.

KaiS's picture

Measuring with music or anything as a stimulus isn't a new thing, it's practically around since we have digital signal processing, say the late' 80s.

The measurement method is called:
Transfer Function, it is an acoustic measurement tool that mathematically compares a reference signal, which can be pink noise or program material, with the same signal after it has gone through a signal chain and speakers / headphones. Since it is looking at the difference between the signals, factors such as room reverberation and noise are minimized.

It isn't very popular for single device measurements such as headphones under a controlled environment, as the results are similar, but the primary effort to do is much greater.
One advantage, once it's set up, measurement results come in much faster, as you don't need longer averaging times for precision results compared to doing measuring with pink noise combined with traditional filtering, or waiting for a sine sweep go through the frequency range.

Would be curious to see if Mr. Foley can add something that is not yet known.

Herb Reichert's picture

The term 'transfer function' (i.e. comparing signal input to signal output) emerged out of filter theory and dates back to at least before WWII. I first read about it the RCA Radiotron Designer's Handbook (4th Edition - 1952) and the "Active-Filter Cookbook - a SAMs publication.

However, today I know of no one using piano notes or solo drum sound to replace the standard sine wave stimulated THD + Noise measurements as Mr. Foley demonstrated in his lecture. One of the problems he outlined alluded to a tester's inability (or adversity to) recognizing and interpreting nuance in a device's output.

I wish I knew more.


KaiS's picture

Lots of signal processing methodies or algorithms were developed long time before digital processing was invented and made them practically usable in the way we do today, e.g. (Fast) Fourier Transformation FFT, first used by Carl Friedrich Gauß in 1805!

Unfortunately I cannot find Dan Foley's mentioned publication, would be interesting to read.

For Transfer Function only two things matter, as it compares the change of an output signal in relationship to the input:

1. Exact delay compensation to make sure only corresponding signals are compared.
2. Signal to noise ratio SNR.

Therefore any stimulus (like music) can be used, but only frequencies that are present in the stimulus by high enough amount can shield sensefull results.
So piano music isn't particularly well suited, a drum solo might be better, or you use a mix of both, hence a normal "band playing music" piece.

This is at least true for frequency and impulse response measurements.
I'm not sure how to, practically, get proper distortion figures out of such kind of music as stimulus, where distortion artifacts are hidden inside the excitation signal.
But - there are proper pulsive signals for this too, e.g. a very fast sine sweep can be used for, but only coarse, distortion measurement - within to the limits of the then low SNR, because the stimulus is short and therefore contains very low energy.

The idea not to use (quasi-)steady sine waves origins from the fact that, especially acoustic and other resonant systems, deliver different results when in steady state opposed to being excited with pulsive signals like music, Pink Noise, MLSSA pulse or other special kinds of short signals.

The goal is to get more practically relevant results, besides other advantages:
e.g. measuring a PA frequency response during a concert using the sound the band produces, with the audience being present.

For headphones I found the differences in measurement results are minor, the masses involved are so small that there are only few resonant effects.

Interestingly the human ear can clearly determine the difference between dynamic, planarmagnetic or electrostatic headphones, where typically the electrostatic has audibly better impulse response.

Simply Nobody's picture

Wonder whether they will be showing RAAL SR1a ribbon headphones, at the show ........ Those headphones can be directly connected to a power-amp/integrated-amp loudspeaker outputs with a supplied adapting device ....... Check out RAAL website for details :-) .........

von Schatu's picture

It would be pretty good if Jude instead of traveling around the globe (I saw him in London last year) and giving lectures about measurement just finally start to acutally bulid up an accessible, consistent and detailed measurement database much like Tyll did.

And no offence, but the measurements of head-fi are inconsistent and useless. I have not seen two measurements produced the same way. Different smoothing, different compensation, etc. pick your poison. When Tyll measured something and I corrected it according to the Harman curve I almost always could identify the measured characteristics with what I have heard. With head-fi measurements it is barely the case. Look no further than the infamous MDR-Z1R case where he "did not detect" any 10k peak in a headphone that when I listened made even a damn Ed Sheeran song sibilant!