Speaker Asylum General speaker questions for audio and home theater.

---

[ Follow Ups ] [ Post Followup ] Thread: [ Display  All  Email ] [ Speaker Asylum ]

---

[ Alert Moderator ]

---

RE: Relevant loudspeaker tests

Posted by tomservo (M) on July 11, 2014 at 08:12:24

In Reply to: Relevant loudspeaker tests posted by BigguyinATL on July 9, 2014 at 10:26:07:

Hi
Dick Heyser spent a good chunk of his life trying to connect or as he put it map what we measure to what we hear and given how our ears and brain work to construct a 3d conscious reality from just two inputs, it has become very clear the microphone and ears work very differently.

A few examples, what our 3d acoustic reality is, is partly a matter of our brain processing the signals such that we discriminate out noise and interference without our knowledge. A look at the HRTF’s or in the ear responses with a measurement eye and one sees something pretty terrible BUT we hear none of those as flaws, instead all those artifacts are how we “hear” a 3d image.

Some years back, co-worker Doug Jones (researching how we hear stereo) made a set of recordings called LEDR which use the pinna responses HRTF’s to “create” the position of a source.

http://www.audiocheck.net/audiotests_ledr.php

Making it more complicated so far as measurements, what sounds good is only vaguely connected to what one measures, for example with music, low number even order harmonic distortion can / usually sounds good when added to the music while higher order / odd harmonics are bad sounding BUT with natural sounds which are not harmonic, either even or odd changes the sound to the negative.
More complications are that we can’t hear distortion if it is very short, Voltage clipping seen on an oscilloscope may be completely inaudible as a flaw but when compared to “without”, the clipped signal sounds less dynamic. To be clear, if clipping lasts long enough or is large enough, it is audible as the flaw we know as clipping.

Even more bewildering is the fact that what we “hear” (as opposed to what sound enters our ears) is partly based on what we know and see. Our senses are intertwined and since that is the only thing we have ever known, it is invisible to us until one faces a situation where one has to depend on one sense alone.

For example, how what we see can dominate what our ears hear can be demonstrated with the McGurke effect, an example of this, try this with headphones and watch carefully;

http://www.youtube.com/watch?v=G-lN8vWm3m0

For example, what we know effects what we hear, again use headphones but unlike the Mcgurke effect, this only works once as once you “know”, you also hear the answer.. Again use headphones and watch carefully / pay close attention when Poppy gives her presentation a few min into the video.

http://www.youtube.com/watch?v=BYTlN6wjcvQ

Now is measurement a lost cause? Heavens no but one has to accept that what sounds good may or may not be the same as accurate reproduction as this is entirely subjective.
Can one measure anything that is an indicator of quality?

Well if one starts with the premise that a loudspeaker should be a faithful reproducer even if the playback material sounds bad, then yes.

Once one has a loudspeaker with flat enough response and few enough other flaws, then the Modulation transfer function (MTF) measurement is somewhat of an indicator.
For example, if one wants to predict how well or how intelligible a voice will be when a loudspeaker is placed in a room, then the STIpa measurement has proven to be a language independent and in Europe legally binding predictor.

The STIpa is composed of 7 MTF measurements in the voice range and voice spectrum but what it is can be more easily envisioned by looking at the optical MTF.

http://www.trioptics.com/knowledgebase/mtf.php

The connection to “what it sounds like” can be made or at least tangentially made to faithful reproduction with a generation loss recording. When we started the company and were pursuing a different kind of loudspeaker, we used these as a reality check and recording both ours and other loudspeakers.

The reasoning was that a loudspeaker should be faithful enough to be included in a generation loss test and should be faithful to the signal enough to survive much more than one generation before sounding bad. We did these on a tower outdoors or in a large warehouse using a good measurement microphone (for some a trusty 4007) at a meter and 24/96 recorder. This way one only heard the loudspeaker itself (and the only thing one can alter in the development process).

One can do this in a room too but the omni mic picks up more room sound than your ears do so the result is HIGHLY dependent on the source directivity and position in the room. Point being unlike any part of the electronic chain, very few loudspeakers can survive 2 or 3 generations and not sound bad. Each generation becomes an exaggeration of what is wrong, what isn’t faithful to the signal and while this doesn’t tell you what’s wrong, it makes where it is much easier to hear and so find.
Funny too, just listening through good headphones in mono to a good measurement mic is revealing in that it by passes your brains sorting of HRTF’s and spatial processing.

If you get used to normal room sounds and other people’s voices first (this is important as a baseline) and THEN listen to one loudspeaker, you can often hear the loudspeakers artifacts which will show up as a sore thumb in a generation or two and essentially invisible to you heard normally.

Anyway, while MTF measurements are obscure and onlt comparative, they do have a casual relationship to how many generations the loudspeaker can pass before “excess audio transmogrification”, in voice reproduction, the STIpa measurement based on MTF’s is an accurate predictor of intelligibility and in stereo, the greater the MTF at the listening position, the more of the direct sound your hearing within the ITG (initial time gap between the direct and then reflected sound) and more palatable the mono phantom or stereo image can be.

Ideally, the sources do not also radiate spatial cues such that when a mono signal is played, you only hear a strong mono phantom and cannot hear a right and left source and this also requires good MTF’s.

Fwiw, the only things one can EQ without messing up phase are minimum phase things and so one need to measure that baseline response ideally outdoors where all you have is the loudspeaker. Once indoors he thumb rule is only cut peaks.

Pink noise is not a good way to measure where to eq as it is blind to phase but is very useful in finding radiated interference patterns as these are audible where you move back and forth and much more so than using music.
On a large scale, you hear the interference pattern on line arrays even with music, as soon as there is any wind and the delayed arrivals usually associated with any interference pattern or comb filtering also lower MTF’s by filling in some of the off time.

Yes, the esl-57 is good, it has directivity and so too the esl63. Both of these even preserve the input wave shape passively and the behavior of the esl-63 was part of the inspiration for the radiation of the speakers at work.
Anyway, this is sort of a different view or my view of Heyser's map that might be of interest. I wish he were still around.
Best,
Tom
Danley Sound Labs

This post is made possible by the generous support of people like you and our sponsors:

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---