Revel Salon2 vs Genelec 8351B - Blind Test Preparations

[HooStat]

You might consider doing things several different ways, just for comparison's sake. Keep your core setup as described above. Then do some comparisons using some suggestions from people for different setups (e.g., a high pass filter and no sub, no sub and no filter, horizontal vs. vertical, mono, etc.). I think your core setup is sufficient to get a good measure of preference and I think it would be interesting to understand how parts of the testing process might have affected the results. With really good speakers, it is more about preferring one over another than good vs. bad. So understanding some room setup keys that affect preferences in a real-world setup would be helpful and interesting. Just my 2 cents.

 

[KSTR]

For Mono vs. Stereo, both have their place here IME, depending on what one is after. For overall spectral coloration and spectral stability vs small listening angles a mono single speaker tends to be more useful (notably with pink noise stimulus), ditto for reverberant field spectral differences, whereas for "soundstage projection" etc stereo is the obvious choice, using more dynamic signals. For distortion again, mono with "organ chord"-style of window-gated triple sines does well, and single steady-state single sine (up to ~2kHz) is good for distortion, and in stereo it shows any minute acoustic asymmetry including the sensitivity to head movement quite brutally (as does mono pink noise played back with stereo speakers).
With real music, it depends on the content, parts of which may correlate to the above (and more) isolated scenarios...

 

[watchnerd]

With really good speakers, it is more about preferring one over another than good vs. bad.

For me, even this can take a long time. I often find myself in a situation where I prefer speaker A on music 1, but speaker B on music 2.

So dialing in a clear preference can take a long time, a hour of listening.

 

[Juhazi]

Why does applying room correction below the Schroeder frequency make the comparison pointless? Even if the filters applied are the same for both speakers?

Because difference between them is practically nulled then. But ok if one accepts that subwoofers must be used with them anyway. Even then I would eq only the sub, then we are left with the difference of woofer layout and height plus distortion above 80Hz. Eq up to say 200Hz will null also these differencies.

What's the point of comparing only mids and tweeters of two very different 3-way speakers?

 

[echopraxia]

Because difference between them is practically nulled then. But ok if one accepts that subwoofers must be used with them anyway. Even then I would eq only the sub, then we are left with the difference of woofer layout and height plus distortion above 80Hz. Eq up to say 200Hz will null also these differencies.

What's the point of comparing only mids and tweeters of two very different 3-way speakers?

Ok, here are measurements with no sub, 1/24 smoothing, 10hz - 22khz:

Revel Salon2:

Genelec 8351B:

What useful info do you think a listening test between these (as such) would conclude?

 

[DonH56]

Would you recommend just a steep high pass filter then, identical for each speaker? No additional correction beyond that necessary, or would it make sense to EQ to an identical LF roll-off response for each of the speakers?

I have not been following this thread and fear I have stepped in the middle, very sorry for the churn.

"Recommend" is too strong -- it depends upon what you are after, why you are testing, and your end goals. If I want to compare midrange, arguable what makes most sense in this case, then yes I would put a HPF somewhere before the power amp(s) so it is applied to both speakers. That way you keep the source material essentially the "same" and do not stress the LF range of the smaller speaker -- assuming in the final system that is how it will work. How high depends upon the speakers, what you are wanting to compare, how you intend to use them in the system, etc. "Same" in the sense that you can play pop/rock, jazz, a symphony, or whatever and not worry about the results being dominated by the extended LF range of the Salon2's.

Again, I am assuming the Genelecs don't reach as deep, something I do not know. I am also assuming you want to see if the speakers are comparable within their nominal performance range rather than going beyond that range. If the Salon2's have an extra octave or two of bass, and you want that, why do the comparison? Or you want that and will use a sub to get there, then either compare the region above the sub, or audition them as they will be used with the sub. It's your test, you get to make the conditions, and can ignore the faceless Internet posters (me).

As far as limiting the response, a somewhat gross analogy would be to compare power amps. If you need 300 W to achieve your target SPL there is not much point in comparing a 300 W amp and a 50 W amp. Maybe a 300W amp and a 500 W amp... But if you have a pair of identical 300 W amps and are debating using them to bi-amp versus picking up a 100 W tweeter amp, then limiting the response to HF only and power levels to those appropriate for the tweeter makes sense. If the HF section will never see more than say 50 W without exceeding tolerable SPL (or damaging the tweeter), then the test may find the higher power of the 300 W amp leads to excessive hiss or whatever.

HTH - Don

 

[Laserjock]

What about suspending the Genelec over the Revel ?
Angle both to ear height.

 

[March Audio]

I have an unfounded theory:

Harman preferred mono testing because that's what their testing rig with the movable platforms could do most easily.

Science history is full of examples of tests and hypotheses fitting the available testing apparatus.

No because they also performed stereo tests and always found the preference order remained the same.

 

[echopraxia]

I have not been following this thread and fear I have stepped in the middle, very sorry for the churn.

"Recommend" is too strong -- it depends upon what you are after, why you are testing, and your end goals. If I want to compare midrange, arguable what makes most sense in this case, then yes I would put a HPF somewhere before the power amp(s) so it is applied to both speakers. That way you keep the source material essentially the "same" and do not stress the LF range of the smaller speaker -- assuming in the final system that is how it will work. How high depends upon the speakers, what you are wanting to compare, how you intend to use them in the system, etc. "Same" in the sense that you can play pop/rock, jazz, a symphony, or whatever and not worry about the results being dominated by the extended LF range of the Salon2's.

Again, I am assuming the Genelecs don't reach as deep, something I do not know. I am also assuming you want to see if the speakers are comparable within their nominal performance range rather than going beyond that range. If the Salon2's have an extra octave or two of bass, and you want that, why do the comparison? Or you want that and will use a sub to get there, then either compare the region above the sub, or audition them as they will be used with the sub. It's your test, you get to make the conditions, and can ignore the faceless Internet posters (me).

As far as limiting the response, a somewhat gross analogy would be to compare power amps. If you need 300 W to achieve your target SPL there is not much point in comparing a 300 W amp and a 50 W amp. Maybe a 300W amp and a 500 W amp... But if you have a pair of identical 300 W amps and are debating using them to bi-amp versus picking up a 100 W tweeter amp, then limiting the response to HF only and power levels to those appropriate for the tweeter makes sense. If the HF section will never see more than say 50 W without exceeding tolerable SPL (or damaging the tweeter), then the test may find the higher power of the 300 W amp leads to excessive hiss or whatever.

HTH - Don

Due to the price range and price difference of these speakers, it just doesn't seem fair to compare without some way of dealing with the bass differences. My objective is to determine which speaker has better sound quality, not just quantity of bass or SPL capabilities. My objective is to determine which speakers yield the ideal high fidelity listening experience, assuming that any listener pursuing such a thing will have subwoofer(s) at least for the Genelecs to ensure flat response spanning 20hz - 22khz.

 

[CDMC]

My thoughts:

• No matter how you do this, people will find an issue with your methodology.
• There is no way for you to do a true double blind or even single blind, where the participants don’t know what they are listening to or comparing two speakers. You don’t own a lab, have a staff, the time, or the resources, so people need to get over it.
• At the end of the day, it seems the goal is to see which speaker people like better without knowing which is which. You might look at the shootout Audioholics did, as they did a good job with the limited resources they had: https://www.audioholics.com/tower-speaker-reviews/1500-2000-floor-shootout-2009
• As you state, the level matching is hard. If you have one speaker that is picked with a strong preference, you may want to reduce the volume of it .5db and run again and see the results.
• Have fun. This should be fun and educational.
• Thanks for doing this.
  
  

 

[watchnerd]

No because they also performed stereo tests and always found the preference order remained the same.

Yes, see remarks above.

 

[watchnerd]

Ok, here are measurements with no sub, 1/24 smoothing, 10hz - 22khz:

Revel Salon2:
View attachment 79308

Genelec 8351B:
View attachment 79309

What useful info do you think a listening test between these (as such) would conclude?

Depends a lot on the music.

Solo acoustic guitar, for example, isn't going to hit the low bass.

Given the music I listen to (lots of jazz, some classical), I might prefer the Genelec even without a sub.

 

[Inner Space]

Yet it is within echoic memory.

The only uncompromised loudspeaker blind test I saw was at a large broadcasting facility, where the testers had use of four identical rooms close to each other. Speakers and LP were set up optimally in each. (One room had a second pair of the same speakers as a control, so there were three DUTs in total.) Participants wore airline sleep masks and were led room to room and chair to chair.

The max interval for the changeovers was less than two minutes. Results were consistent and led to a large purchase that folks were happy with as long as I was there. Yet, technically, the echoic memory limit had been exceeded, which worried the testers. But on review, it was clear we have rather thoughtlessly imported some superficially similar research into an arena where it really doesn't belong.

Echoic memory research - with its well known four- or six-second limit - is about cognitive or intellectual retrieval of meaning from an ongoing barrage of information. The quoted limit is about the "read back" capability - i.e. how long can you recall, and therefore re-examine, part of an argument before it is swamped by new incoming parts of the argument. Close to our use case, perhaps, but not exactly targeted, and not exactly pertinent.

I think we suffer from patchy, old, threadbare and incomplete data points. A lot of what we rely on dates from telephony and early radio, and kinda-sorta similar parallel research from other fields. I think we should always be ready to question what we think we know, and certainly we should re-examine the so-called four-second limit for audio memory, which I think would be an early casualty of further research.

 

[echopraxia]

This is interesting. Genelec 8251B bass appers to be lower distortion than Revel Salon2 (at least until it starts reaching its extreme limits)!

Revel Salon2:

Genelec 8351B:

However, I can't push the Genelec beyond this without the limiter engaging. And I'm a little afraid to push the Salon2 much further as well, given how much excursion I'm seeing from the woofers at this level at 20hz.

Here is the distortion measurement from the Genelec when the limited engages in the low frequency portion of the sweep:

If I high pass these both, I'm sure both will be able to reach excessive SPL easily. But these bass limits is also why I like using subwoofers, even with the Salon2's: it just improves headroom and allows fearlessly pushing very high SPL down to 10hz. This applies to both speakers here.

Edit: Actually, it looks like the Salon2's have higher distortion in the 2nd and 3rd harmonic, but much less in 4th 5th and 6th, shown here:

Revel Salon2 - 4th, 5th, 6th distortion harmonics:

Genelec 8351B - 4th, 5th, 6th distortion harmonics:

Not sure how much of this if any would be audible though.

 

[ctrl]

how do you deal with switching back and forth rapidly when the listener wants to go over a part of the song repeatedly on each speaker? I’m not sure how Harman structures the test procedure, but in many cases this ability for participants to switch back and forth quickly seems to be helpful and desirable.

Fast back and forth switching is not possible, but it should be faster than the speaker change in the Harman tests.

I‘m getting a lot of mixed signals here. Some say it’s useless in mono, others say it’s useless in stereo.

This is my biggest criticism of Toole and Olive's research, who have developed a sound evaluation algorithm for free-standing loudspeakers with mono signals.
Many people who start with loudspeaker development make the mistake of tuning their first self-developed loudspeaker individually, standing freely in the room, with a mono signal - speak from experience
If both loudspeakers are then listened to in stereo in a normal listening room, they sound much too bright.

Would even go so far as to say that if a speaker standing freely in the room sounds optimal with a mono signal, it's a crossover mismatch.

Anyone can easily check this by placing a speaker freely and adjusting a mono signal with an equalizer so that it sounds optimal (at about 85dB) at the listening position and then listening to the stereo setting.

 

[DonH56]

Most speakers seem to design for flat frequency response. Your assertion implies a single speaker should include a HF roll-off to compensate for typical rooms. That is not how most I have seen are designed, and since rooms are so variable it makes more sense to me to design the loudspeaker for flat response and let the user, and his/her room, adjust to taste.

In any event I have to get back to work (yes, it's Saturday) and step away from this land-mine of a thread. - Don

 

[Laserjock]

Most speakers seem to design for flat frequency response. Your assertion implies a single speaker should include a HF roll-off to compensate for typical rooms. That is not how most I have seen are designed, and since rooms are so variable it makes more sense to me to design the loudspeaker for flat response and let the user, and his/her room, adjust to taste.

In any event I have to get back to work (yes, it's Saturday) and step away from this land-mine of a thread. - Don

Work in a Saturday? What’s that?

 

[watchnerd]

If both loudspeakers are then listened to in stereo in a normal listening room, they sound much too bright.

Well, I've certainly found some Harman speakers to be brighter than I prefer, but I don't know if that's from the research impacting speaker design or the "house sound" of JBL.

Or just my head transfer function / pinnae.

 

[LTig]

I have followed this thread and don't understand the criticism against using room EQ below Schroeder. We all know that frequency response below Schroder heavily depends on the room which has a much bigger influence on bass quality than the speaker's FR. This is the reason why @amirm always uses a fixed EQ for his subjective listening tests of speakers.

Especially in this case where @echopraxia wants to find out which speaker sounds better in his room, using room EQ is the best approach - he would use room EQ anyway later, wouldn't he?

 

[preload]

I am not going to be a primary listener here. I expect at least 3 untrained listeners, and hopefully more, none of which include myself. I will participate but only out of curiosity, and will identify the logged results that I produce as belonging to my test session so anyone reading the results can filter the data however you like.

Ohhhh okay then it all makes sense then. I had similar thoughts to @Beave. If you already know two speakers well, you're going to be able to identify them in a blind ABX. And knowing the identity defeats the purpose of a blind experiment because it reintroduces the bias you were trying to control for. You may want to mention the 3 unfamiliar listener part in your op.

 

[preload]

So my last blind test of Revel F206 vs Ascend Sierra RAAL Towers was almost meaningless because trivially correctable bass differences of the Revel made the comparison much less interesting (a tie, where Revel F206 won bass and Ascend won treble). In general, bass differences will determine the winner, in my experience, especially if there are no major flaws in the rest.

This is why I would worry about running both full range like that with no bass normalization. But I am certainly open to the idea, especially for these — because both of these very nearly true full range speakers.

In order to remove the influence of differences in bass capability, did you consider just doing the abx with same subwoofer for both the genelec and the revel and crossing them over at, say, 80 hz? I'm assuming you don't want to penalize the genelec for not being a floorstander. Then you wouldn't need to worry about eq (at least eq in the bass region) since it would be the same sub in both comparators.