John Atkinson's of Stereophile Talks About Measurements

[audiophile]

Boy, this is an excellent example. Subjective reviewers like it with their eyes open, while Stereophile's measurements indicate that it's broken by design.

I agree, an excellent example, because Border Patrol DAC was preferred to a Benchmark by most listeners in a blind test too:
https://www.stereophile.com/content...nalogue-converter-se-jon-iveson-november-2018

 

[xr100]

Well, of course there's a "normal" range of circumstance and then there's needing hearing aids right?

Sure.

I did have a hearing test myself a few years ago (actually as part of a checkup related to Bell's Palsy!)--it's within the "normal range"--but I do have a dip around 200Hz or so in my right ear. No, it's not "corrected" for in my systems.

However... well, I have yet to respond to #135. Briefly, though, I don't see any problem with "shelving" type corrections, and, if desired, "loudness" type controls (and I don't mean the on/off loudness buttons found on pre/integrated amplifiers!) and upward compression of low frequencies can be useful at lower listening levels. What needs to be avoided (to prevent serious colouration) is peaky response behaviour etc.

It is interesting to "sweep" bandpass filters around when listening to recorded material. It can be surprising to hear frequency bands where certain instruments unexpectedly "drop out." It's a sign that the technique of "subtractive EQ" is going on, attenuating parts of an instrument's spectrum that do more harm than good by being in the same frequency range. (aka Critical bands, masking, etc.) Also, that's what acoustic musical instruments are up to, as well--resonances/spectral character is effectively "EQ" to get a type of instrument heard above other types, and the size of the instrument causes LF roll-off.

In a way, it's all one big "hearing aid" for normal humans. Of course, not done carefully, it's dreadful (as in radio station "maximisation" etc.)

 

[Blumlein 88]

I agree, an excellent example, because Border Patrol DAC was preferred to a Benchmark by most listeners in a blind test too:
https://www.stereophile.com/content...nalogue-converter-se-jon-iveson-november-2018

Okay two things jump out.

Firstly, oh god , please will everyone get the memo that level matching with SPL meters is not precise enough. Just stop it. If you cannot get and use a multimeter to set levels properly you are incompetent to continue. JA should know this.

Two, frequency response differences are readily audible via pink noise, and that seems to have been the case here. So not surprising they have a different sound.

Now where is the fine detailed data that most listenenrs preferred the Border Patrol? Some did, some in the first part of the blind test said the BP was better on poorly recorded material and not on well recorded material. Seems some leaps of logic involved here.

And yes like in the article, I'd be perfectly okay with a Benchmark and a plug in or process using DSP to make it sound like the BP if that is someone's preference.

 

[MRC01]

... What, then, in principle, is the difference between using this and shelving up the HF with EQ? How does one know whether a recording is neutral or not? What does that even mean? ...

If live musical performance is the absolute reference, it's a moving target. It sounds different depending on the room, where in the room are the musicians and listeners, and different listeners in the same performance hear a different sound. Because of this, the definition of "neutral" covers a wide range of different sounding recordings. So as a listener I try to be flexible in what I consider to be a "neutral" recording -- or a "good" or "bad" recording.

 

[xr100]

If live musical performance is the absolute reference

It can only be a reference in the sense that if you went all out trying to "accurately" capture an acoustically-generated performance and then "accurately" "reproduce" (the perception of) it, then the two could be compared (albeit caveated by the extent to which this is possible, certainly not by 2 channel stereo.)

It is useful to listen to the reproduction of certain sources, such as the human voice to check for colouration, as we are so sensitive.

However, in general, is it a reference for the vast majority of recorded material? Is all-acoustically generated sound by definition the best possible quality?

The latter, i.e. the goal of generating the best possible sound, and this being provided by attempting to "accurately recreate" acoustically generated events, is what has become conflated in "hi-fi." (And, for one, allows for a great deal of nonsense to be written in "subjectivist" reviews.)

So as a listener I try to be flexible in what I consider to be a "neutral" recording -- or a "good" or "bad" recording.

Despite Floyd Toole's "Circle of Confusion," I find most recordings that I listen to are actually quite well balanced. But then they're the ones I chose, so--hopefully--they would tend to be? I'm not about to cut any slack if they aren't... and there are plenty of examples of what I would consider to be bad sounding instruments and overall poor sound quality provided by a recording. And, with music, at least, it's getting worse with the changes in the recording industry. At least movies still get to be mixed at "Skywalker Ranch."

Recordings by a voiceover artist captured using the Manley Reference Cardioid, Manley Reference Gold, Bock 195, MXL Revelation, and Telefunken U47: (I seem to recall that J. Gordon Holt recorded a microphone test that was on a Stereophile test disc to illustrate the difference in sound. But I never had a copy...)

 

[snapsc]

A few comments:

1. I suspect the Fletcher Munson curves somewhat explain why a downwards sloping 20 to 20 curve is preferred in order to get the proper balance

2. Since most of us are not "trained" listeners...I'd like to know how what a trained listener hears and prefers helps to predict what the untrained listener is going to prefer....consider me untrained and more of the "I'll know if I like it when I hear it" type

3. As to recordings to demo with...I prefer music that I really like, that is well recorded (doesn't have to be audiophile perfect) and that I am very familiar with....it just makes it easier to hear differences and sort out preferences when other gear is changes

 

[xr100]

A few comments:

1. I suspect the Fletcher Munson curves somewhat explain why a downwards sloping 20 to 20 curve is preferred in order to get the proper balance

What is the typical spectral balance of recordings...?

In most pop/rock/EDM/etc., usually, the highest amplitude sound is the kick. If it's not been brickwall limited, then the waveform of the track will "spike" around loud percussive hits.

A basic way of "modelling" a kick is a transient high amplitude burst of "noise" at the start, giving way to a sine(ish) wave which drops from maybe 200 to 50Hz over, say, 150-200ms, gradually losing energy (amplitude as well as the drop in frequency.) Hmm, so which part of the frequency spectrum makes it perceptually loud? The bottom end will start to be physically "felt" as well as "heard." But if you do a narrow boost down there, then you've ended up messing with the "envelope" of the instrument, i.e. raising the amplitude where the kick is "supposed" to be losing energy.

It's not that straightforward...

If you listen to recordings that have a spectral balance that looks "boosted" in the top-end, they will tend to have a harsh sound.

It's interesting to use low pass, band pass and high pass filters to divide up the spectrum. This can easily be done with a (3 way or 2 way plus sub) active system by turning on one "band" at a time. If you listen to a sub on its own, or a tweeter on its own, the perception doesn't quite relate to the full bandwidth one. It's certainly not as loud as you'd expect. Switch everything back on, with the midrange doing its thing, and it all comes back together again. Perception "just works" without conscious intervention or awareness so we don't realise how much processing is going on in our heads. And, the frequency range chosen for telephone systems wasn't an accident, that's the most important area.

2. Since most of us are not "trained" listeners...I'd like to know how what a trained listener hears and prefers helps to predict what the untrained listener is going to prefer...

I, too, would be interested in this query being elaborated upon.

that is well recorded (doesn't have to be audiophile perfect)

What is this "audiophile perfect" recording and where can I find it?

 

[MattHooper]

Given that (presumably--I haven't systematically checked!) there is absolutely no consistency between "subjective" reviews in one publication and another, one might as well just write the review using a "random review generator," modified up or down based upon who bought the drinks at the last trade show... even then very similar products within a manufacturer's line have ended up with widely different reviews in the same publication!

I understand why someone would presume that to be the case, but from my own observations I don't see the situation as that dire. I've seen some good convergence in subjective descriptions in the sound of a number of speakers from reviewers. And in having been part of audio forums since the 90's, I've found there can often be quite a bit of convergence and agreement on the general nature of various speakers. There's much more divergence as to who likes them or not, but careful listeners who are decent at putting sound in to description can often enough converge pretty well.

I used to do a bit of reviewing and have some pals who still do it. When one of my writer friends gets in a new speaker review I always drop over to listen. It's not uncommon for us to diverge on our feelings about the speaker in question - he tends to like far more of them than I do - but in discussion the actual sonic character we seem to be 'hearing and describing' essentially the same thing.

 

[xr100]

I understand why someone would presume that to be the case, but from my own observations I don't see the situation as that dire.

You're right, it wasn't terribly balanced--though it was supposed to be somewhat tongue-in-cheek...

In fairness, as well as acquiring false ideas and thinking that later had to be dispensed with or modified, I learnt a great deal from the "hi-fi magazines" back in the day--certainly basics from "don't buy a mini system from the local electronics emporium, it will suck", to frequency response, impedance, and so on. And, dare I say, the even the more "flowery-language" reviews at least got one thinking about listening critically and provided some sort of tentative vocabulary to use.

And if you read something like the DIY Supplement of Hi-Fi World magazine, it would even have reviews of technical books--they were always recommending "The Art of Electronics." (Yes, the one by Horowitz and Hill.)

I've seen some good convergence in subjective descriptions in the sound of a number of speakers from reviewers.

I rarely read "subjective" (really for want of a better word) reviews anymore. When I did, in the 1990's, I'm pretty sure that one magazine would declare a certain product the "amplifier of the year" or "CD player of the year" and, seemingly eons later, it would be reviewed in another magazine that gave a distinctly middling view. The fact that the manufacturer of said "product of the year" always placed prominently positioned full-page adverts in the magazine couldn't possibly have anything to do with it....?

There were certainly products that were "seminal" or substantially different enough that (my distant recollection of) the reviews was that they were more consistent.

In the mid-1990's, the Meridian 565 (or Lexicon DC-1) were the only games in town for what they did. And, obviously, one could point to a relatively unique product like the Quad Electrostatics.

In the "hi-fi press" as a whole, though, the vast majority of reviews were of commodity "me too" products that weren't distinguished in any way. The magazines that bothered to perform measurements (and publish them) would sometimes "dismiss" a product based on very bad engineering.

One particularly good example was a review of the Mission 751's in Hi-Fi Choice. It didn't sound very good, and the measured performance looked "odd." In a follow-up, it was said that Mission had subsequently "discovered" the supplier of the bass/mid driver had changed its parameters, or at least the bass/mid drivers in the review sample were way out of spec. Mission said that they had resolved the problem and invited Hi-Fi Choice to pick a random sample from their warehouse. Which, apparently, they did, and subsequently re-reviewed the product. Quite what that says about Mission's QA process at the time, well...

Another nice example was the Sony CDP-715E CD player, which Hi-Fi Choice found had the best measured performance of any CD player they'd reviewed--they gave it the highest recommendation. It retailed for the "beer budget" (as Paul Messenger would say) price of £200. (Of course one could be cynical and wonder if there hadn't been a squabble with one of Sony's competitors, but AFAIK Sony really did make some of the best delta-sigma DAC chips then.)

Perhaps it's worth mentioning that Hi-Fi Choice did--and maybe they still do--publish "group reviews" that made use of blind tests with a panel of listeners. Albeit, I'm not sure how well controlled they were, in terms of protocol (not double-blind?), level matching, etc.; and quite often whoever wrote up the tests would "overrule" the panel with their own opinion!

OK, enough waffling on about distant memories--LOL. Yes, as I said, I think you're right that a more balanced appraisal would be fair. But these days I'd go for an engineering approach to the whole system. For instance, with loudspeakers I'd first want to see the polar response and so on (and the fact that the typical bass/mid with separated dome tweeter design is not too good in that respect cuts out a large chunk of the products on the market!)

I used to do a bit of reviewing and have some pals who still do it.

I'm not sure if this is good or bad, but in fairness to yourself, I don't think I've ever read any of your reviews.

And yes, thanks to the Internet at large, including forums, it's down to the individual to seek out good information out of the embarrassment of riches scattered amongst the weeds. Where would we be without it?!

 

[Nathan_A]

I loved the comment of liking something that’s wrong. I fear that I fall in that camp with my love of my primaluna tube amp and vinyl setup. I’ve been following these forums for a long time but I’m hesitant to take the plunge into the accuracy world. I love what I have and although it may be wrong, I still love it.. what if I get it right, but I still prefer what’s wrong?

My way of dealing with this is to get accuracy from my components and then introduce the wrongness I like on purpose and variably on top of that.

I recently picked up a mint condition Yamaha MX-1, snagged a Yamaha C-85 for cheap (until I can track down a CX-1 for sale) and feed it with a Topping D50s. This is all replacing a Pioneer SC-65 reveiver that I'm moving elsewhere.

My room is bright. Lots of hardwood flooring, glass, etc. So all the high frequencies are really a problem. I can't do room treatments because this is all in the main and only open living space. My speakers are ONIX Reference 3's which I used to drive years ago with an Aragon 8008B and an ONIX CD-1 player (MBL knock off) in a really stereo friendly listening environment. So, I have a fair sense what everything is "supposed" to sound like.

All the vintage Yamaha gear is at least as good as any of that old setup, and definitely more enjoyable than the AVR (though I have to say, for an AVR it's pretty impressive). The problem is the listening environment itself.

So... when I'm listening to EDM music or something where I don't want harsh highs-freq. reflections to drown out the bass and mid-bass... I just roll them off with the preamp EQ. When I'm listening to a really poorly produced pop song with the bass boosted, then I tame that with the EQ. When I'm listening to accoustical music where nothing is happening in frequencies that are really problematic in my environment, then I just defeat the EQ.

It took me a very, very long time to accept this as a way of handling my listening preferences. I must have spent north of $50k over my lifetime trying to EQ my music through the characteristics of different components.

In no small part because of ASR, I've decided to just get components that measure well, and change the sound of my music whenever I want, and however I want, on a case by case basis through the adjustment of a couple knobs instead of through the depletion of my bank account.

 

[MattHooper]

So... when I'm listening to EDM music or something where I don't want harsh highs-freq. reflections to drown out the bass and mid-bass... I just roll them off with the preamp EQ. When I'm listening to a really poorly produced pop song with the bass boosted, then I tame that with the EQ. When I'm listening to accoustical music where nothing is happening in frequencies that are really problematic in my environment, then I just defeat the EQ.

It took me a very, very long time to accept this as a way of handling my listening preferences. I must have spent north of $50k over my lifetime trying to EQ my music through the characteristics of different components.

In no small part because of ASR, I've decided to just get components that measure well, and change the sound of my music whenever I want, and however I want, on a case by case basis through the adjustment of a couple knobs instead of through the depletion of my bank account.

That's good stuff!

I happen to have landed in a sort of similar place, though from the opposite side as it were.

To the degree I have wanted to "color" the sound of my system it's generally been through using some older tube amplification. To the degree they may be coloring the sound, it seems I simply like it across the board. If anything it's quite subtle, but I like it. It doesn't remotely homogonize things in term of allowing large or small differences between tracks come through. It's been essentially a "set-and-forget" situation as the amps have powered my system for over 20 years. I never have an urge to twiddle with EQ.

In fact until recently I'd had a digital EQ sitting in my system since the late 90's. I just kept it in there thinking "someday I'll need it, someday I'll feel the need." But I had to eventually face the fact I never really used it and never felt the need. So I sold it a while back.

I enjoy the differences in production quality/choices. It's interesting to me. So I don't ever put on a piece of music with the mind that "maybe I'll want to eq this to sound substantially different." If there's little bass or whatever in a track, I'm ok with that. I like putting on music without having to contemplate how I might want to change it.

(Now...that's not *entirely* true because I do play with my room acoustics here and there....)

 

[audiophile]

To the degree they may be coloring the sound, it seems I simply like it across the board. If anything it's quite subtle, but I like it. It doesn't remotely homogonize things in term of allowing large or small differences between tracks come through. It's been essentially a "set-and-forget" situation as the amps have powered my system for over 20 years. I never have an urge to twiddle with EQ.

I am in a similar situation. My tube amp and r2r DAC produce pleasant natural sound, which works well for me on all records. I only use EQ on subwoofers to compensate for variations in the amount of low-frequencies between records from the 1980s and modern music, which can be up to 8 dB.

 

[RayDunzl]

what does an un-colored pink noise spectrum look like? Is it flat or is it sloped?

Using REW:

With RTA display pink is flat, white slopes up
With Spectrum display pink is sloped down, white is flat.

The story was that he was playing (violin) in the DC Metro and no-one stopped to listen.

Somebody did... At 1:35 in the video.

 

[dshreter]

Using REW:

With RTA display pink is flat, white slopes up
With Spectrum display pink is sloped down, white is flat.

What is the difference between RTA and spectrum display? I'm interested to understand this better, especially if there are any choice links that go into this topic, and the relative merits of each for assessing actual in-room system performance.

 

[vavan]

What is the difference between RTA and spectrum display? I'm interested to understand this better, especially if there are any choice links that go into this topic, and the relative merits of each for assessing actual in-room system performance.

did you read https://www.roomeqwizard.com/help/help_en-GB/html/spectrum.html ?

 

[DonH56]

REW made RTA mode look more like an old audio analyzer and uses octave weighting (bins) to show the data. Spectral view has the same data but puts it into evenly-spaced bins like a "normal" FFT. The difference is how the spectra are displayed.

Since pink noise has equal energy per octave, it displays "flat" in RTA mode since RTA mode is per octave, and slopes down in spectrum mode since the energy falls off as you go up in frequency. Pink noise falls off with frequency -- there are more FFT bins per octave since the octaves span more frequencies if you want to think of it that way. 20 Hz to 40 Hz is one octave, and 10 kHz to 20 kHz is one octave, but if your FFT bin is 20 Hz wide there is one in the first example and many in the second. Since white noise is "flat" with frequency, it displays flat in spectrum mode, but rises in RTA mode since there are more "Hz" in each octave as you go up so more energy as frequency increases.

HTH - Don

 

[Sgt. Ear Ache]

Using REW:

With RTA display pink is flat, white slopes up
With Spectrum display pink is sloped down, white is flat.
.

yes I know. I wasn't actually asking...was more a rhetorical question in response to the previous discussion

 

[xr100]

Since pink noise has equal energy per octave, it displays "flat" in RTA mode since RTA mode is per octave, and slopes down in spectrum mode since the energy falls off as you go up in frequency.

Have not used REW... surely the "slope" in "spectrum" mode can be set?

 

[DonH56]

Have not used REW... surely the "slope" in "spectrum" mode can be set?

View attachment 47663View attachment 47664

I don't actually use REW all that often myself so do not know. It's not much like the RF spectrum analyzer at work...

 

[xr100]

I don't actually use REW all that often myself so do not know. It's not much like the RF spectrum analyzer at work...

Lacking by comparison?