Genelec 8341A SAM™ Studio Monitor Review

[QMuse]

The interesting question is where the border between those two, when is it good enough to be not audible and when it that border exceeded.

Any peak in the FR is some kind of resonance. If you are speaker designer than it is up to you to find the root and correct it. You may do it with some dumpening, reinforcing or you can apply EQ based on precise (pseudo) anechoic measurement.

But once you bought the speaker all you can do is room EQ, make the whiskey glass test and don't crank them to the limit where concentric waves in the glass will start to form. In that case drink some more whiskey and crank them more as chances are you won't notice it.

 

[thewas]

But once you bought the speaker all you can do is room EQ

Not necessarily only, for example I had improved some 80s Japanese design loudspeakers which my girlfriend wanted by adding alubutyl sheets and extra damping inside, but as you say most wouldn't do so.

 

[QMuse]

Not necessarily only, for example I had improved some 80s Japanese design loudspeakers which my girlfriend wanted by adding alubutyl sheets and extra damping inside, but as you say most wouldn't do so.

Nice!

Btw, how did you measured the improvement? Or was it not important as thankful smile of your girlfriend was rewarding enough?

 

[andreasmaaan]

I dont think there's really much dynamic EQ happening. There is some weirdness happening at the very highest volume setting (100 percent in devialet's app), but all the other ones track seem to track each other very closely above 60Hz. Compression only really seems to affect the region below 60Hz. I quite appreciate this detail since usually it begins higher up into the mids and even mid-bass. The way compression is built into the Reactor means you still get some bass oomph throughout most of the bass, it's just the sub-bass that might be limited.

Actually, I've been using the term "dynamic EQ" the way you've been using the term "compression" anyway. It's really just a semantic difference TBH, both have the same effect.

In the meantime, here's the Reactor 900 vs the Kef R3 in room. I've moved some things around in my room since the R3 measurements were made but nothing that should really affect the low bass. Note the R3 had the benefit of having both speakers pretty much perfectly aligned with one another timing wise.

Interesting comparison, thanks

Obviously this isn't the ideal way to compare in-room measurements, but I think it should still give you an idea of how much room gain I'm working with (and that the Reactors do indeed have an unnecessary boost in the bass).

I'm not sure this is the case TBH. If you look at this measurement you took:

IMO it's hard to say whether the apparent boost below 120-ish Hz reflects the anechoic behaviour of the Reactor, or whether it's an artefact resulting from the nearfield measurement technique. When you take a nearfield measurement, there's a tendency for the effect of baffle step to be obscured (in effect, reduced) as a result of the close proximity of the microphone to the driver. This can theoretically result in up to 6dB of apparent gain below the baffle step frequency range, although usually less in practice. So it may be that if you were able to take a true anechoic measurement of the Reactor, it would show that it's actually flat in the bass.

The way I estimated your room gain at 10-12dB at that mic location was by looking at the two graphs you posted and making some assumptions (all subject to some uncertainty ofc):

• the Reactor has a basically flat frequency response from 25Hz to 10+ kHz (based on the reasoning I just explained)
• room gain will on average be greater at lower frequencies, and will tend towards zero in the mid/high frequencies
• SPL in your room reduces by approximately 3dB with each doubling of distance (this is based on experimental results from other rooms, see for example Fig. 4 below, taken from this paper)

From this starting point, we might look at the 80dB in-room measurement you took of two reactors at a distance of 3m and infer that each reactor was producing a bit above 80dB anechoic:

• 80*80dB = 86dB for two correlated sound sources
• -5(ish)dB for an increase in distance by a factor of 3
• = 81(ish) dB

The room gain in your room can then be estimated by looking at the difference between a predicted flat response 25Hz-10kHz (the Reactor measured anechoically) and the measured response of the Reactor in your room:

We can also then infer (loosely) that the Reactor's dynamic EQ/bass compressor begins kicking in at some point between when 81(ish) and 86(ish) dB (estimated anechoic).

That's my loose reasoning, anyway.

Interestingly, it seems to match pretty closely with what the Bagby spreadsheet predicts.

 

[andreasmaaan]

I'm aware of all this - this is room measurement and that's how things go there. Besies, I didn't post that measurement for a direct comparison with free space measurement but for the folks to try to measure themselves with REW. All rooms will have similar gain so those figures will be comparable. As I said, I doubt small speakers can reach the same SPL with the same THD at the same position and same distance. Have you tried yours?

The midrange drivers in my current setup are 12" - no small speakers in the house here

 

[QMuse]

Actually, I've been using the term "dynamic EQ" the way you've been using the term "compression" anyway. It's really just a semantic difference TBH, both have the same effect.

IMHO difference between those 2 exists. One is deliberate action of the built-in SW logic to save speaker from excessive excursion and related distortion (dynamic EQ) and the other is a passive side effect of not doing so (compression). And I don't think they have the same effect - similar yes, but definitely not the same.

 

[QMuse]

The midrange drivers in my current setup are 12" - no small speakers in the house here

But it's the cabinet size that matters as well, right? Remember our beloved Genelec which has 2 woofers packed in a small enclosure?

Anyway, measure, post here and receive either compliments or boos!

 

[andreasmaaan]

IMHO difference between those 2 exists. One is deliberate action of the built-in SW logic to save speaker from excessive excursion and related distortion (dynamic EQ) and the other is a passive side effect of not doing so (compression). And I don't think they have the same effect - similar yes, but definitely not the same.

Compression has different meanings in different contexts. You're referring to power compression, which is a physical effect resulting from a woofer being overdriven. That of course is very different from what is happening here.

I was referring to (multiband) dynamic compression, which is functionally the same as dynamic EQ.

 

[QMuse]

Compression has different meanings in different contexts. You're referring to power compression, which is a physical effect resulting from a woofer being overdriven. That of course is very different from what is happening here.

I was referring to (multiband) dynamic compression, which is functionally the same as dynamic EQ.

I was actually referring to "dynamic comrpession" term used by @napilopez. Did he use that term referring also to an artifical active process (action of electronic/software) as dynamic EQ is?

 

[napilopez]

Actually, I've been using the term "dynamic EQ" the way you've been using the term "compression" anyway. It's really just a semantic difference TBH, both have the same effect.



Interesting comparison, thanks



I'm not sure this is the case TBH. If you look at this measurement you took:

IMO it's hard to say whether the apparent boost below 120-ish Hz reflects the anechoic behaviour of the Reactor, or whether it's an artefact resulting from the nearfield measurement technique. When you take a nearfield measurement, there's a tendency for the effect of baffle step to be obscured (in effect, reduced) as a result of the close proximity of the microphone to the driver. This can theoretically result in up to 6dB of apparent gain below the baffle step frequency range, although usually less in practice. So it may be that if you were able to take a true anechoic measurement of the Reactor, it would show that it's actually flat in the bass.

The way I estimated your room gain at 10-12dB at that mic location was by looking at the two graphs you posted and making some assumptions (all subject to some uncertainty ofc):

• the Reactor has a basically flat frequency response from 25Hz to 10+ kHz (based on the reasoning I just explained)
• room gain will on average be greater at lower frequencies, and will tend towards zero in the mid/high frequencies
• SPL in your room reduces by approximately 3dB with each doubling of distance (this is based on experimental results from other rooms, see for example Fig. 4 below, taken from this paper)

View attachment 51660

From this starting point, we might look at the 80dB in-room measurement you took of two reactors at a distance of 3m and infer that each reactor was producing a bit above 80dB anechoic:

• 80*80dB = 86dB for two correlated sound sources
• -5(ish)dB for an increase in distance by a factor of 3
• = 81(ish) dB

The room gain in your room can then be estimated by looking at the difference between a predicted flat response 25Hz-10kHz (the Reactor measured anechoically) and the measured response of the Reactor in your room:

View attachment 51662

We can also then infer (loosely) that the Reactor's dynamic EQ/bass compressor begins kicking in at some point between when 81(ish) and 86(ish) dB (estimated anechoic).

That's my loose reasoning, anyway.

Interestingly, it seems to match pretty closely with what the Bagby spreadsheet predicts.

On Dynamic EQ, understood! When you said Dynamic EQ I was thinking of loudness EQ - as in with the purpose to compensate for equal loudness countouts rather than to protect the drivers.

I also appreciate your help with correlating in-room SPL levels to anechoic! However, my measurement of the reactor has already been compensated for baffle step. This is also true of every other spliced-bass quasi-anechoic measurement I've posted on ASR. So what you see in the splice is the bass level after the reduction. The non-compensated nearfield bass is several dB higher.

I use the same Bagby spreadsheet to model the baffle step. So far, this has correlated quite closely with other anechoic measurements, and though obviously there's always going to be some room for error, I'm still quite confident that bump in the bass is inherent to the speaker.

Anyway, since Devialet seems to have made a few fixes since my review and I have a few more questions, I'm going to see if so can get them to send me a speaker again (maybe I can convince them to send Amir one too). I have access to my building's roof so while not practical for most measurements, I can use that to try to calibrate my quasi anechoic measurements to something closer to freefield conditions.

 

[andreasmaaan]

I was actually referring to "dynamic comrpession" term used by @napilopez. Did he use that term referring also to an artifical active process (action of electronic/software) as dynamic EQ is?

The cause of confusion simply arises because all that we've been discussing relates to compression of a signal of one kind or another. The type of compression really just comes down to questions of whether the effect is intentional/electronic (in this case digital) or unintentional/electrical.

Generally speaking, the former is referred to as dynamic compression (or dynamic range compression), and the latter as power compression.

 

[andreasmaaan]

On Dynamic EQ, understood! When you said Dynamic EQ I was thinking of loudness EQ - as in with the purpose to compensate for equal loudness countouts rather than to protect the drivers.

Yes, I think I caused this confusion to some extent myself with my earlier post in which I misread your measurements.

I also appreciate your help with correlating in-room SPL levels to anechoic! However, my measurement of the reactor has already been compensated for baffle loss. This is also true of every other spliced-bass quasi-anechoic measurement I've posted on ASR. So What you see in the splice is the bass level after the reduction. The non-compensated nearfield bass is several dB higher.

I use the same Bagby spreadsheet to model the baffle step. So far, this has correlated quite closely with other anechoic measurements, and though obviously there's always going to be some room for error, I'm still quite confident that bump in the bass is inherent to the speaker.

Understood Sorry, I shouldn't have assumed you hadn't taken this into account.

This means your measured room gain correlates even more closely with Bagby's calculator, in that case.

 

[andreasmaaan]

Using the same method as with room EQ - only Q<=1 are allowed.

Could you elaborate or link to a more detailed explanation?

 

[QMuse]

The cause of confusion simply arises because all that we've been discussing relates to compression of a signal of one kind or another. The type of compression really just comes down to questions of whether the effect is intentional/electronic (in this case digital) or unintentional/electrical.

Generally speaking, the former is referred to as dynamic compression (or dynamic range compression), and the latter as power compression.

I see, so dynamic compression = dynamic EQ?

 

[andreasmaaan]

I see, so dynamic compression = dynamic EQ?

Um, the two are functionally equivalent, but the interfaces on typical effects unit are different. Dynamic EQ gives more precise control than (multiband) dynamic compression. The former allows for parametric-type dynamic EQ, while the latter offers only action on specified frequency bands.

There's a discussion of the differences here:

The Difference between Dynamic EQ and Multiband Compression

Dynamic equalizers and multiband compressors are similar in certain key respects: both compress and expand selectively; both process your audio in a way that link the dynamics of a track with its frequency spectrum.

But whereas multiband compressors use crossover filters, which affect fairly broad frequency areas, a dynamic EQ allows you to specify the precise frequencies you want to boost (through expansion) or attenuate (through compression). The F6 Floating-Band Dynamic EQ offers six such “floating,” or fully adjustable bands, each with full control over width (“Q”), gain, range, threshold, attack, and release. What’s more, internal and external sidechain inputs are provided for each band so that the specified frequency is compressed or expanded only when the sidechain signal exceeds a user-defined threshold. In addition, a unique “split/wide” mode allows you to optionally filter the audio being used as the sidechain trigger.

All these features make dynamic equalization a powerful and extremely flexible tool that can be applied during mixing or mastering, as well as in live performance. By letting you surgically zero in on trouble spots and then treat them only when trouble arises, a dynamic EQ can serve as the ultimate problem solver.

 

[QMuse]

Could you elaborate or link to a more detailed explanation?

Sure, I can elaborate. Step response pre-ringing depends on phase filters Q factor, frequency and ammount of correction. As soon as you define a single phase filter within all-pass filter you would have pre-ringing, the question is only how much is acceptable (non-audible). It turns out that few mild phase filters, say less than 60deg, with Q=1 are not audible, but more than that will cause audible pre-ringing.

To demonstrate this I defined 3 filters of 45deg at 100 and 1000Hz, with Q=1, 2 and 3.

Here is how it looks:

Phase with 100Hz filters:

Step with 100Hz filters:

Phase with 1000Hz filters:

Step with 1000Hz filters:

So, staying with Q<=1 and <=60 deg you are on the safe side. But start using filters with higher Q and with more deg correction things will soon get very nasty and they will become audible.

Keep also in mind this pre-ringing is much more audibleble than artifacts you see in GD response. which look nasty but are actually very benign.

 

[andreasmaaan]

Sure I can elaborate. Step response pre-ringing depends on phase filters Q factor, frequency and amount of correction. As soon as you define a single phase filter within all-pass filter you would have pre-ringing, the question is only how much is acceptable (non-audible). It turns out that few mild phase filters, say less than 60deg, with Q=1 are not audible, but more than that will cause audible pre-ringing.

To demonstrate this I defined 3 filters of 45deg at 100 and 1000Hz, with Q=1, 2 and 3.

Here is how it looks:

Phase with 100Hz filters:

View attachment 51668

Step with 100Hz filters:

View attachment 51669

Phase with 1000Hz filters:

View attachment 51670

Step with 1000Hz filters:

View attachment 51671

So, staying with Q<=1 and <=60 deg you are on the safe side. But start using filters with higher Q and with more deg correction things will soon get very nasty and they will become audible.

Keep also in mind this pre-ringing is much more audibleble than artifacts you see in GD response. which look nasty but are actually very benign.

Ah yes thanks I agree with that. The issue is that Devialet doesn't claim no audible pre-echo, they claim:

""The process is strictly causal and generates no pre-echo."

The filters you're describing are not strictly causal, and do generate (likely inaudible) pre-echo.

 

[QMuse]

And this is what you get if you have nasty cabinet resonance..

Nope, just kidding, this is what you get after phase correcting LR24 passive crossover. May not look very nice but not an audible pre-ringing.

 

[QMuse]

Ah yes thanks I agree with that. The issue is that Devialet doesn't claim no audible pre-echo, they claim:

""The process is strictly causal and generates no pre-echo."

The filters you're describing are not strictly causal, and do generate (likely inaudible) pre-echo.

Aha. Well, I would certainly like to know how they are doing it, but unfortunately I don't. Please share if you ever find out!

EDIT: Some room eq solutions are also doing phase correction without introducing any pre-ringing, probably using the same trick..

 

[thewas]

Nice!

Btw, how did you measured the improvement? Or was it not important as thankful smile of your girlfriend was rewarding enough?

At that time I unfortunately didn't do any comparison measurements before I did those improvements also on the 2nd loudspeaker, hope one day I will find a third one (as I wouldn't want to rip the existing ones apart) and do then some measurements and blinded ABX tests with them.