Shootout between JBL M2 and Revel Salon 2

[NorthSky]

Classical Opera music = dynamic. ...With the full orchestra backing up the vocals. ...The power, the glory and the emotional fury, from a whisper to a thunder.
Motion Picture Soundtracks = very dynamic → http://www.complex.com/pop-culture/2013/09/epic-movie-scores/sunshine

 

[hvbias]

One more example of a film soundtrack. When there is some average level dialog going on and there is a sudden explosion that releases a lot of bass energy into the room. It's dynamic because it is transient and occurring at a significantly louder volume than the rest of the dialog that preceded it. Movies often get criticized for having audio tracks that are too dynamic, where low level information gets harder to hear because of the loud dynamic parts which leads to people constantly reaching for their remotes turning the sound up and down. This is mitigated to some effect with proper level adjustments to individual speakers.

======================

Now my subjective opinion- I don't think we need horn speakers for music where we are at the very most seeing 18-20 dB of dynamic range with the most intense symphony music. Amplifier power is cheap and even the most inefficient (ok maybe not something like Apogees) speakers should never be clipping.

I would love to see a Toole run blind test to see if the poor decay (ringing) in the treble of compression drivers was aggravating to listeners in a blind test. Based on my own experience with headphones that had poor treble decay on CSD, this was most definitely audible. Maybe it makes less difference with speakers in a real room where there are many other interactions.

 

[NorthSky]

You want dynamic music?

 

[oivavoi]

Now my subjective opinion- I don't think we need horn speakers for music where we are at the very most seeing 18-20 dB of dynamic range with the most intense symphony music. Amplifier power is cheap and even the most inefficient (ok maybe not something like Apogees) speakers should never be clipping.

I've been wondering whether that 18-20 dynamic range approximation is correct, though. As far as I've understood it, the apps that are used for creating the DR range database for example are employing some smoothing, which means that very brief and highly dynamic transients probably won't show up. My hunch is that the actual, non-smoothed DR on well-recorded stuff probably is more like 30 Db. Not a problem on most studio stuff nowadays, but could be noticeable on actual acoustic recordings.

Another question which is very under-explored, is dynamic homogeneity. When transients occur with lots of information going on at the same time, is the driver able to reproduce all of the frequencies with the same dynamics? If a trumpet shrieks and a bass drum kicks in at the same time, do the frequencies follow the exact and correct dynamic signal? This might be difficult to investigate with steady state test tones. I believe Sweden's Ingmar Öhman has concerned himself with this issue quite a bit.

To horn or not to horn is only one of the issues - the other is whether there's a point in buying amps with much more power than one thinks one needs.

 

[Jakob1863]

I've been wondering whether that 18-20 dynamic range approximation is correct, though. As far as I've understood it, the apps that are used for creating the DR range database for example are employing some smoothing, which means that very brief and highly dynamic transients probably won't show up. My hunch is that the actual, non-smoothed DR on well-recorded stuff probably is more like 30 Db. Not a problem on most studio stuff nowadays, but could be noticeable on actual acoustic recordings.

Another question which is very under-explored, is dynamic homogeneity. When transients occur with lots of information going on at the same time, is the driver able to reproduce all of the frequencies with the same dynamics? If a trumpet shrieks and a bass drum kicks in at the same time, do the frequencies follow the exact and correct dynamic signal? This might be difficult to investigate with steady state test tones. I believe Sweden's Ingmar Öhman has concerned himself with this issue quite a bit.

To horn or not to horn is only one of the issues - the other is whether there's a point in buying amps with much more power than one thinks one needs.

Normally the dynamic range would denote the difference between the lowest level and the highest level of music signals in a track, but in fact the DR value is calculated in a differenct way:

http://www.dynamicrange.de/sites/default/files/Measuring DR ENv3.pdf

so it reflects more the loudness of the signal.

 

[oivavoi]

Normally the dynamic range would denote the difference between the lowest level and the highest level of music signals in a track, but in fact the DR value is calculated in a differenct way:

http://www.dynamicrange.de/sites/default/files/Measuring DR ENv3.pdf

so it reflects more the loudness of the signal.

Interesting. Thanks!

 

[Blumlein 88]

You want dynamic music?

Yeah, but you'll have to put up with the near continuous high pitched tone playing in some parts of the soundtrack.

 

[watchnerd]

Interesting. Goes against my expectations. The interesting question then becomes: why were the Salons preferred?

It's exactly what I expected. Most listeners prefer the cushier sounds of hi-fi speakers to the stiff reality check of monitors.

 

[Blumlein 88]

If you haven't read it, Floyd Toole's summation is worth reading. Posted today.
http://www.avsforum.com/forum/89-sp...iewed-speakers-ever-made-19.html#post54644120

 

[watchnerd]

If you haven't read it, Floyd Toole's summation is worth reading. Posted today.
http://www.avsforum.com/forum/89-sp...iewed-speakers-ever-made-19.html#post54644120

I was disappointed that it was all general stuff, with little about the specific test or speakers at hand.

 

[A.wayne]

It's exactly what I expected. Most listeners prefer the cushier sounds of hi-fi speakers to the stiff reality check of monitors.

Maybe they preferred it because the stiff reality of the monitors were just bad ....



Just saying ...

 

[Blumlein 88]

The question I would like answered is according to the spin o Rama results there's an 86% correlation with which speaker listeners will prefer. So does it predict a win by the M2 or the Salons?

 

[watchnerd]

Maybe they preferred it because the stiff reality of the monitors were just bad ....



Just saying ...

Or mis-EQed.

But I'd be hard pressed to say the M2 are bad given the general accolades.

 

[watchnerd]

The question I would like answered is according to the spin o Rama results there's an 86% correlation with which speaker listeners will prefer. So does it predict a win by the M2 or the Salons?

Really?

Not doubting, but would love to read where that was documented.

 

[RayDunzl]

So does it predict a win by the M2 or the Salons?

Yes.

 

[watchnerd]

I'm curious how they'd do sighted, because I think the M2 looks butch Darth Vader and the Salon 2 looks fussy and prissy.

 

[Blumlein 88]

Really?

Not doubting, but would love to read where that was documented.

I might can dig it up. I think Floyd Tool had a link somewhere in that long thread.

Yes, however, over time they have developed a predictive model, and claim in general they can correlate the spin-o-rama results to listener preference 86%.

Now if you know some basic statistics that isn't quite what many people think. It doesn't mean 86% accuracy for instance. It simply means as spin-o-rama results improve listener preference is correlated with it and accounts for about 86% of the increased listener preference. So what I don't know is how do they judge spinorama results. What is a result that is twice as good or half as good. This is a strong correlation of listener preference to measured performance. It is quite an achievement in this field of speaker design.

So we know various levels of listener training have less variability and more precision in such tests. We don't know how this group compared. The results were generally very close as well. It is possible if you do the test again it turns out the other way. It is clear both of these speakers were ranked quite highly. Still, my assumption was the M2 had the best spin-o-rama results of any of their products. With some of the EQ disabled on the M2 should that still have been the case?

 

[Blumlein 88]

I concurred with some comments in the AVS thread about panel speakers. Someone said they preferred panels in stereo, but found them poor and uninvolving in mono. I agree. I have owned 7 different panel speakers over the years. I have considerable time spent listening to about twice that many owned by friends. I've the curious experience that mono recordings over a pair of panels is great. A mono recording over one panel is not so great. A really good box speaker in mono using only one speaker seems much better.

Now it makes me wonder about the Harman testing of panels in mono. Yet I am not sure where it should stand. It is also true that most panels have rather poor frequency response. You would expect that to hurt it in mono. The directivity issues and dipole nature also get swallowed up some when you mix all that from two sources in ways it will not listening to one.

Then I have this experience. At one time I put one Soundlab on one channel, and one well regarded box speaker on the other channel. Used my Tact room correction to run the test signals and develop correction curves just that way. A wildly different curve of correction between the channels. The correction was good enough you heard nothing obviously amiss. Imaging, which I thought would suffer badly, instead sounded fine. It did not quite seem like either speaker, but it wasn't all messed up or wavering up and down the frequency band. It actually was one of those experiences that kept pricking my audiophile mind about how highly discriminating (or not) our hearing was. The room correction impressed me, and impressed upon me that your hearing wasn't that hard to fool. That keeps me skeptical when people who know more than me tell me it isn't a good idea to correct above the Schroeder frequency. I actually tried that at the time. Without correction above 500 hz it was just about the mismatched mess you would imagine.

So I sound like an audiophool this way, but I wonder if there isn't something about panel dipoles in mono that alter that spin-o-rama vs listener preference correlation curve. Yet the little data I have read about seems like the measured panel results of strange directivity and uneven FR predict just about how poorly they do in the blind testing. My experience correcting panels indicates a greatly improved listening experience yet I can't correct for the directivity issues. Plus the several Harman speakers I have heard since getting cued into their design method have simply out-performed anything close at all to their price. And as a lifelong panel guy, I don't miss the panels. The panels still have attributes I prefer, but on balance Harman seems to be onto something important.

 

[Blumlein 88]

Really?

Not doubting, but would love to read where that was documented.

The best source might be to purchase the new edition of Toole's book.
https://www.amazon.com/Sound-Reprod...p/B008VSMXNK/ref=mt_kindle?_encoding=UTF8&me=


Second best is even more expensive. The standards for measuring home loudspeakers. Which basically is from Harman's work I take it.

https://www.amazon.com/Standard-Method-Measurement-Loudspeakers-Formerly/dp/B01LYK70TR

Here is the table of contents for that book.

https://standards.cta.tech/kwspub/published_docs/ANSI-CTA-2034-A-Preview.pdf

Best short discussion of it I could find at the moment:
http://www.audioholics.com/loudspeaker-design/measure-loudspeaker-performance

The 86% number was from testing 70 loudspeakers and the results as of 2004. One hopes maybe they have improved upon that.

If you don't know what the spin-o-rama is, it is measurements in an anechoic chamber at 10 degree intervals vertically around the speaker. And at 10 degree intervals horizontally around the speaker.

Given the results, and the target is flat on axis and very evenly sloping response just off axis it supports the idea frequency response is 85% of audio as I say quite often.

 

[Jakob1863]

Logan et al. gave a nice presentation wrt new loudspeaker measurement methods:

http://www.warkwyn.com/wp-content/uploads/2015/01/NFS-ALMA-2015-PPT.pdf

Olive even had his multiregession analysis method patented and there are two AES references:

Olive, Sean E., "A Multiple Regression Model for Predicting Loudspeaker Preference Using Objective Measurements: Part I--Listening Test Results," Audio Engineering Society 116th Convention, Berlin, Germany, May 8-11, 2004

Olive, Sean E., "A Multiple Regression Model for Predicting Loudspeaker Preference Using Objective Measurements: Part II--Development of the Model," Audio Engineering Society 117th Convention, San Francisco, CA, October 28-31, 2004

in which he compared his model to the consumer report model and claimed his much better correlation coeffecient.

Edit: link repair....