JBL 4349 Review (Studio Monitor Speaker)

[H-713]

Generally, for pleasure listening, dips in frequency response are less objectionable than peaks. A big peak in that region would be far more unpleasant.

I have a feeling that they really, really wanted to use a 15" woofer (or perhaps that's what marketing told the engineers they had to use!), and they did the best they could without DSP. It could be worse in this regard- I don't see any evidence of cone breakup on the woofer in the frequency response, so they've done their homework there.

It's possible to do very good work on some "objectively broken" studio monitors. Strangely enough, some of the monitors that people like the most (and have the best success with) are far from being ruler-flat objectively perfect speakers. As such, anyone wanting to use these as monitors would have to try them. The dynamic performance does bode well.

One question for Amir- what distance were these measured at? At least in my experience, some larger speakers like this look absolutely horrific until you're a couple meters away (which is how these are likely to be used). Does the Klippel do this (or have some way of calculating it)?

 

[Sancus]

@amirm PIR graph seems to be missing?

 

[polmuaddib]

With all due respect, this review feels more subjective then objective.
I admit I am not knowledgeble enough, I read only third of Floyd Toole's book (some parts require rereading for me), but I can't escape the feeling that there is a lot of bias toward the brand and price and sensitivity of the speaker.
And can someone show me measurement for dynamics? And explain it, because I keep thinking of dynamic range, but that isn't the same, right?
Because, wherever I read online, dynamics is defined differently and loosely.
Thanks

 

[amirm]

@amirm PIR graph seems to be missing?

Oops. Kept thinking something was missing. Added it to the review:

 

[napilopez]

By the way, here's the one harman spin I could find, which is heavily smoothed (via Speakerdata2034)

Via the comparator, here is how it compares to amir's measurement,with the SPLs as plotted (just LW and ER):

So mostly very similar, except Harman's doesn't show the bright trendline/shelfed highs.

Scaling the harman measurement down by 1.5dB to better line up in the bass, we see the response there matches well:

 

[JohnBooty]

I can certainly see some flak coming for liking and recommending a speaker that doesn't measure well, Mr Harman dealer

@amirm can you please elaborate a bit more about why you liked it (especially since you smoothed out the tonality with eq?) I think that might help those of us better assess how to approach speakers that don't measure well

This is more expensive than the 708p, and measures terribly.

Feel like I really read a different review than you guys, or maybe you're just focusing on the preference score. These measure "terribly?" What I'm seeing is good (but not perfect) FR and directivity measurements coupled with some very nice low distortion numbers.

The practical reality... in terms of subjective listener enjoyment... seems to be that low THD really matters, especially the challenging 96dB distortion test which is a good indicator of how cleanly and effortlessly the speakers reproduce dynamic peaks. Excellent performance in this regard compensates for less than perfect performance in other regards.

Low distortion is great, but most speakers already go plenty loud.

Most speakers do not do 96dB like this. There's a big difference between doing 96dB and doing it cleanly.

All in all, Amir's measurements and subjective impressions agree with what a lot of people have subjectively reported over the years, which is that ability to cleanly reproduce big dynamic peaks is a major contributor to an enjoyable listening experience. A major weakness of the Harman formula seems to be that it does not take this into account.

This matches my experience. I've got JBL 3 Series monitors as my daily drivers. Going by the "preference score" they are more or less world class and would embarrass these 4349s. However in reality dynamic range is an obvious limitation of the 3 Series monitors relative to big boys like these 4349s.

 

[amirm]

I admit I am not knowledgeble enough, I read only third of Floyd Toole's book (some parts require rereading for me), but I can't escape the feeling that there is a lot of bias toward the brand and price and sensitivity of the speaker.

I accept the appearance of such. As I noted in the review, I was ready to give this speaker a very harsh rating. Listening tests showed not only the benefits of dynamics, but also the relative harmlessness of crossover dip, and ease with which EQ could fix it. Pyschoacoustics backs this in the way our hearing bandwidth shrinks as frequencies go up.

But sure. You can assume as much fault as you like in my subjective assessment. You have the measurements and that is all objective with no ability on my part to influence them. Heck, I could have changed the switches and measured again to remove the shelving of the tweeter but did not do that.

As I have explained before, I can't lie about my subjective impressions. Whatever they are, you are going to hear them. If they always agreed with measurements, then I would have never started any commentary to go with my measurements!

 

[thewas]

I like it. I think I read these measurements differently than some of you.

The way I see it, the speaker is excellent except for one significant flaw in the FR in the directivity. The flaw is a dip, as opposed to a peak, and is relatively narrow, so I suspect it's not as audible as it may look. But 200-1000Hz and 2kHz to 10kHz is unusallly smooth for a passive speaker.

Completely anecdotally, I very rarely find issues between 1-2kHz -- where so many speakers have issues, to be all that detrimental. This is especially true if the problem is a dip. I have no smart reason for this, but it's something I keep on noticing. I notice anomalies in almost every other frequency range so much more... 2-10kHz is particularly crucial for me.

In terms of the preference score, it suffers primarily from the nearly constant directivity-ish design as indicated by the SM_PIR. Honestly, I suspect I'm more likely to be bothered by the fact that the speaker's trendline seems to be a bit bright. So I'd definitely use that HF shelf.

Is it worth $7,000 n measurements alone? Probably not, and I wouldn't recommend it based on that alone given some of the other options. But if you listen to it and like it and appreciate the design, I don't see this as a dealbreaker frequency response.

I also agree and find it well engineered for a large passive old school 2-way design, if I would want to nitpick I would mention only the a bit too hot tweeter (but thankfully it has dip switches) and the ugly frontal BR ports which have also too much output around 1kHz which could be audible being levelwise too close to the drivers output.

The ASR measurements also match the ones from this review also quite well:

(2nd and 3rd harmonic distortions at 95 dB)

Source: https://www.hifitest.de/test/lautsprecher_stereo/jbl-4349_20054

 

[Sancus]

I accept the appearance of such. As I noted in the review, I was ready to give this speaker a very harsh rating. Listening tests showed not only the benefits of dynamics, but also the relative harmlessness of crossover dip, and ease with which EQ could fix it. Pyschoacoustics backs this in the way our hearing bandwidth shrinks as frequencies go up.

But sure. You can assume as much fault as you like in my subjective assessment. You have the measurements and that is all objective with no ability on my part to influence them. Heck, I could have changed the switches and measured again to remove the shelving of the tweeter but did not do that.

E: You did note the part about the switches in your post, my bad. I will say there are many reviews where you did not change such settings where available and did not give the benefit of any EQ above room mode fixes. So I think it's easy to read in a little bit of bias here

But I will admit that there seems to be something about the ability to play very loud and cleanly in the 20-500hz area that makes people more forgiving about the other flaws of a speaker.

It would be very interesting to do a real blind test of a big speaker like this with poor measurements and a small one like the Genelec 8030C which have scores that would indicate the small speaker will be preferred >95% of the time. Though there might need to be a third speaker involved to avoid people being able to easily recognize which is which.

In general I'm not really that fussed about the subjective part of the reviews and I wish people would take them a little less seriously

 

[martijn86]

This was super pleasurable but less refined

That's the main reason I bought the Klipsch Forte. I had chased perfection for so long that I just had to let loose for a while. What I'd ideally have is one Genelec 8341 + MiniDSP set and one set with large horn speakers to just rock out.
Maybe more people can relate to this: Sometimes I want to sit down for a brilliant whisky, other times I just want to drink cheap rum with friends.

 

[Easternlethal]

I accept the appearance of such. As I noted in the review, I was ready to give this speaker a very harsh rating. Listening tests showed not only the benefits of dynamics, but also the relative harmlessness of crossover dip, and ease with which EQ could fix it. Pyschoacoustics backs this in the way our hearing bandwidth shrinks as frequencies go up.

Feel like I really read a different review than you guys, or maybe you're just focusing on the preference score. These measure "terribly?" What I'm seeing is good (but not perfect) FR and directivity measurements coupled with some very nice low distortion numbers.

The practical reality... in terms of subjective listener enjoyment... seems to be that low THD really matters, especially the challenging 96dB distortion test which is a good indicator of how cleanly and effortlessly the speakers reproduce dynamic peaks. Excellent performance in this regard compensates for less than perfect performance in other regards.

okay so it's the dynamics which compensates. Why doesn't this show up in the measurements? Or does it?

 

[Sancus]

okay so it's the dynamics which compensates. Why doesn't this show up in the measurements? Or does it?

It shows up in the distortion measurements. This speaker has lower distortion from 20-500hz than most of the things that have been measured. The reviews really don't do a good job of examining dynamics though, largely because comprehensively testing that is painful and time consuming. Erin used to do a very nice multitone max SPL stress test but he stopped because it took hours and was sensitive to being interrupted by external noise.

Additionally, the interpretation of these kinds of tests is not very well settled, and there isn't a good standardized way to display distortion that accurately reflects what we would or would not hear. Various solutions have been proposed. The tweeter distortion in this review is a good example. You will never reach 96dB at those frequencies no matter how loud you play, so the distortion numbers in that graph are not useful.

Currently it's just a limitation of the testing that we have to accept.

 

[Vict0r]

Thanks for the review! Would you say there is a clear indication of why the measurements and subjective listening test don't line up? Have you measured speakers where it was the other way around? Like, they measured well but sounded off.

 

[amirm]

Thanks for the review! Would you say there is a clear indication of why the measurements and subjective listening test don't line up?

They line up. Crossover error is the only deviation in on-axis which was fixable by EQ and was too narrow to be audible.

Have you measured speakers where it was the other way around? Like, they measured well but sounded off.

I have. SVS Ultra comes to mind.

 

[JohnBooty]

okay so it's the dynamics which compensates. Why doesn't this show up in the measurements? Or does it?

It's there in the distortion measurements, specifically the 96dB distortion tests. While not a perfect representation of real world usage, they seem to me to be a fairly solid proxy for how well a speaker will hold itself together when pushed and how cleanly it reproduces those dynamic peaks.

Best way to understand the strengths and limitations of the "preference score" is to go buy a set of JBL 305's. According to the preference score they should be all we ever need. They actually are great speakers, but push them and you'll see why we don't all just buy JBL 305's and call it a day.

 

[DualTriode]

Is this higher distortion an artifact of compression drivers in general? The "Voice Coil" reviews of almost all compression drivers seem to indicate high 2nd order distortion is typical. Higher-order distortion is typically much lower and also, while 1-2% 2nd order distortion seems the norm at the measurement spl, it does not seem to increase with higher SPL in the same progression as most dome tweeters would.

I had felt it may be an artifact of the measurement technique used by Voice Coil. But with this review, I am wondering if it's a design choice?
Maybe the ear finds HF 2nd order overtones pleasing? Scanspeak is a very highly rated driver manufacturer and some of their top tweeters measured similarly, higher 2nd order (around -45dB) and then really low higher-order distortion (around -60dB or better). Seems to be mirrored by these measurements.

I have a love hate thing going on with Compression Drivers and horns/wave guides.

One of the biggest things that current technology horns do well I call CD; controlled dispersion and dynamic response is also pretty cool too.

The physics and theory of horns is a lot to chew on. Horns are acoustic impedance transformers. Along the acoustic length there are changes of geometry that have a large impact on the impedance of the horn/waveguide. As a result horns have signature sound, they sound like horns. I have plotted the impedance curves of many horns/waveguides. It is common for CD Horns/waveguides to have three impedance peaks dependent on the horn/waveguide that is attached to the Compression Driver.

Yes CD’s have a family of harmonic distortions that are specific to Compression Drivers. Back behind the CD diaphragm is a small air volume that compresses with the movement of the diaphragm. Air is a nonlinear spring. With compression of the air space the sound output is distorted, the greater the compression the greater the distortion.

I prefer the sound of cones and domes up to ~1.6kHz to 2kHz, above that the Controlled Dispersion is much more important.

 

[TimVG]

Comparison of on-axis and -10°v

 

[Maiky76]

Hi,

Here is my take on the EQ.

The raw data with corrected ER and PIR:

Score no EQ: 4.2
With Sub: 5.9

Spinorama with no EQ:

• Not as Flat as i would expect
• Nice directivity
• Ugly trough

Directivity:
Better stay at tweeter height or just above (but the curve is then significantly different).

EQ design:
I have generated two EQs. The APO config files are attached.

• The first one, labelled, LW is targeted at making the LW flat
• The second, labelled Score, starts with the first one and adds the score as an optimization variable.
• The EQs are designed in the context of regular stereo use i.e. domestic environment, no warranty is provided for a near field use in a studio environment although the LW might be better suited for this purpose.
• We don't see a dramatic rise in THD towards very LF so the High-Pass filter may be converted into a pure boost

Score EQ LW: 6.0
with sub: 7.6

Score EQ Score: 6.4
with sub: 7.9

JBL 4349 APO EQ LW 96000Hz
April212021-160259

Preamp: -3 dB

Filter 1: ON HPQ Fc 38.6 Hz Gain 0 dB Q 1.25
Filter 2: ON PK Fc 106 Hz Gain -3.22 dB Q 1.53
Filter 3: ON PK Fc 239 Hz Gain -2 dB Q 5.8
Filter 4: ON PK Fc 991 Hz Gain -1.45 dB Q 5.67
Filter 5: ON PK Fc 1494 Hz Gain 4.23 dB Q 2.14
Filter 6: ON PK Fc 1922 Hz Gain -3.3 dB Q 2.63
Filter 7: ON PK Fc 3911 Hz Gain -1.31 dB Q 2
Filter 8: ON PK Fc 8310 Hz Gain -2.6 dB Q 2.1

JBL 4349 APO EQ Score 96000Hz
April212021-155811

Preamp: -2.6 dB

Filter 1: ON HPQ Fc 38.6 Hz Gain 0 dB Q 1.25
Filter 2: ON PK Fc 108.5 Hz Gain -3.35 dB Q 1.59
Filter 3: ON PK Fc 239 Hz Gain -2.38 dB Q 7
Filter 4: ON PK Fc 980 Hz Gain -1.58 dB Q 8.1
Filter 5: ON PK Fc 1487 Hz Gain 4.34 dB Q 2.91
Filter 6: ON PK Fc 1912 Hz Gain -3.11 dB Q 1.85
Filter 7: ON PK Fc 3900 Hz Gain -1.77 dB Q 2
Filter 8: ON PK Fc 8358 Hz Gain -3.53 dB Q 1.5

Spinorama EQ LW

Spinorama EQ Score

Zoom PIR-LW-ON

Regression - Tonal

Radar no EQ vs EQ score
Nice improvements

The rest of the plots is attached.

 

[Absolute]

In the M2 at least, the design axis is the logo between the drivers. I'd assume it was the same here with 4349. I guess they're designed to stand on small stands that are a little bit tilted backwards, which makes sense looking at your plot @TimVG .

Interesting to see that the tweeter has the same peaking around 8 kHz as the D2340k in the M2. Horrible for those of us that hate sibilants, but mighty impressive on certain instruments - and fantastic for the sense of dynamics.

 

[TimVG]

When you cross a 12" woofer at 1,5kHz to a large(ish) horn (and accompanying CTC distance) laws of physics dictate a narrow and finicky vertical dispersion pattern. The peaking axis of a driver is usually at (or close to) its physical center. We can see in the normalized vertical directivity that different peaks and dips correlate with the respective drivers as one moves along the vertical line. The exact correlation with audible impressions in this regard seems inconclusive at the moment.

The spinorama would certainly look better if -10°v would be implemented as the on-axis. It's possible through the NFS software to recalculate the curves, if Amir is interested in that. Personally I don't really care, but better looking curves would match the audible impression and limit arguments about the two not going hand in hand.