JBL 4309 Review (Speaker)

[sarumbear]

To be honest, for me, I do not see a difference with speakers that have the headless panter. All these graphs all look the similar bad. Time for me to leave ASR because I have the feeling this is just a Harman promotion forum.

If you don’t value the tests then I wish you good luck and say: bye and don’t let the door hit you.

 

[ROOSKIE]

Well this is deff turning out as an interesting review.
I have now heard quite a few speakers with Harman scores and generally they align resonably well for me. Not an exact correlation however.
Along with experiencing 2 speakers with similar scores sounding very similar, I have deff used two speakers with similar scores that sound substantially different. And speakers with the same score that I prefer much differently.
I have also had at least 1 highish score that I absolutely did not like, ELAC DBR62.
I have also found that for high playback levels most of the data here goes out the window and some speakers just scale up and up while others simply do not sound good loud.
I have also found something that will make some very staunch objectivists cringe, that some speakers truly have that "je ne sais quoi". They just do.

Now return to the real topic here and compare speakers in controlled listening conditions and ask about preference.

As far as I know JBL still blind tests all these speakers before release, so that fact that it exists means it hopefully passed blind preference testing. They certainly have the resources to produce whatever sound they want.
Amirs testing is fairly controlled for a brief sighted test compared with some others in that is under consistent conditions.
He liked the 4349 as well. It is possible that something about this design strategy is really working. Certainly the ability to stay clean and go loud is working.
Anyway I am curious to find out more.

 

[Morpheus]

At first, I could not believe it, but I did a quick sim of the baffle step:

View attachment 158556
That seems to be pretty much spot on. It's a bit shifted to the low side, but that's just a guestimate of the woofer placement:

What about the big waveguide lip the woofer sees especially on the vertcal axis isn't that exacerbating this issues?

 

[GXAlan]

As far as I know JBL still blind tests all these speakers before release, so that fact that it exists means it hopefully passed blind preference testing. They certainly have the resources to produce whatever sound they want.

That’s right.

Currently, it is my understanding that the majority of these monitor type speakers are very popular in Japan. There was a period where these speakers were only available in Japan before coming back as the JBL Synthesis product line.

The Array 1400 that was tested in Floyd O’Toole’s book isn’t as good as the Revel Salon2 but it was really really good still. There are certain advantages to compression drivers (efficiency) and the new generation HDI horn is even better than the classic designs. JBL also seems to like the sound of large 15” and 12” drivers.

I am sure this is slightly tuned for a JBL house sound the same way that JBL headphones and AKG headphones follow the science on what sounds good or popular but allows for a slight tuning of the sound.

Just like the L100, part of the look of the blue baffle monitor speaker is the “big box” speaker. Clearly you would have less diffraction with narrower enclosures but these are typically not sold to blind customers.

 

[ctrl]

Well, the spinorama captures all these things except for the "loud" part. It captures directivity/room interaction and bandwidth.

I would like to agree with that. Without measurements, one should be cautious about making predictions about how loud or dynamic a loudspeaker can play without "significantly" (this has yet to be defined) changing its timbre - i.e. dynamic compression.

Small 2-way speakers are always dynamically limited in the low frequency range. The short-time dynamic measurements (at 76, 86, 96, 102 dB) of @hardisj are very informative, because they introduce an objective measure for the dynamic capability.

To get an idea of the dynamic compression of the JBL 4309, I simply compared the frequency responses of the HD measurements at 86dB and 96dB.

Since these are most likely performed without changing the position of the microphone, it can be assumed that repeated measurements should result in almost identical FRs - or let's assume it was and let's further assume that the amplifier voltage was meticulously adjusted to 10dB difference .

Around 0.5dB sound pressure level difference I would define as dynamic compression, since such a level change, with a correspondingly small Q, is perceptible as a tonality change - this is simply my definition to be able to compare loudspeakers.

(I made small sound pressure level adjustments by hand, e.g. +-0.02dB)

Thus, the dynamic limit of the JBL 4309 would already be clearly exceeded in the low bass around 40Hz. The human voice range starts at about 100Hz, around 120-180Hz the 4309 reaches a deviation of 0.5dB and would thus also be dynamically at the limit in this frequency range.


In comparison, the ASR Directiva, which by this objective standard (as I said, assuming Amir's measurements are comparable) has less dynamic compression and actually only sags a bit in the frequency range above 6kHz. However, it also just reaches the limit of dynamic compression at 40Hz.

 

[heflys20]

To be honest, for me, I do not see a difference with speakers that have the headless panter. All these graphs all look the similar bad. Time for me to leave ASR because I have the feeling this is just a Harman promotion forum.

Well, if that's the case, they're doing a terrible job, since he has disliked other Harman products. The JBL 4319 for instance:

https://www.audiosciencereview.com/forum/index.php?threads/jbl-4319-studio-monitor-review.12986/

 

[Inner Space]

Look at all the people intensely enjoying music from anything: $2 earbuds; inbuilt laptop speakers; bog-standard car cassette tape audio units in the 80's.

Which proves my point - technical parameters, while per se interesting to you, don't affect enjoyment.

Now return to the real topic here and compare speakers in controlled listening conditions and ask about preference.

Which is precisely what Harman did, thereby coming up with the 4309. It was compared blind and preferred. Apparently its ugly FR wasn't a handicap. You need to update your thinking. The evidence in front of you is plain.

 

[beefkabob]

Also this points to the importance of *gasp* subjective impressions?

The question was never only what produces the best measurements. The methodology is that measurements show accuracy and, when on the Harman curve for in-room performance, interact with the room to produce the most pleasing results. All the best research is based on subjective impressions by listeners, often trained, in controlled listening tests.

The danger that this site helps avoid is getting sucked into the eternal upgrade cycle. A world class headphone stack is now relatively cheap. A great speaker, amp, preamp, DAC stack can be done for 3-4k, and much less if you're willing to use powered speakers. Want to go down to 20hz? Then it costs a bit more as you need a great sub.

 

[ROOSKIE]

Very cool to see.
This looks great to me (for both speaker sets)

Bear in mind the Directiva is an active speaker and use the 6.5" woofer that is supposed to be one of the most state of the art in DIY existence and would be very unlikely to make a trip to your home in a $2k speaker, + needing a Pr and what ever else I'd guess $3k+(or even much more) for a typical passive with that kind of woofer/PR combo.
In that thread $4-5k or more was established as a potential fair market price for a finished active with that configuration.

That compression in the low bass is excellent for both 6.5" woofers. Even the 4309 is only 1db down at 40hrz. Seriously pretty good and since the driver/box parameters do change at high volume it may not even be driver compression. It could be the port in the case of the JBL and PR in the case of the Directiva. (which is still an issue but I like knowing the driver may be handling it just fine)
I seriously doubt you could even remotely hear the compression at all in either speaker anywhere in the frequency range@96db. Especially as as bass gets louder we perceive a non linear increase.
It would be interesting to see higher SPL's.
I am impressed with both systems.
Over at EAC many speakers are far worse, or at best similar - the 4309 is roughly on par with something like the Focal Twin6 Be which is active and uses 2 woofers and costs $4000 USD a pair or the KEF R5 small floorstander.
The two recent Klipsch test there with huge woofers did no better.
The JBL M2 and the JBL 708p seemed to represent the possible SOTA on this type of compression. (more speaker tests needed to declare that)

In any case, I still make the case that some speakers sound excellent at high volumes and some do not. I do like to listen loudly sometimes. I do not think these small deviations and HD levels can fully explain why some speakers do loud well (and I mean get loud and sound great doing/ where you still want to go louder because it is so good & fun, not where you are just showing off high SPL for high SPL sake).
Obviously I love to be able to test this stuff blind. Be super fun. I have no way.

I do have the ability to take reasonably meticulous (in room) near field sweep measurements and examine the frequency responses for changes. When I get a chance I will do this for the 4309.

I would like to agree with that. Without measurements, one should be cautious about making predictions about how loud or dynamic a loudspeaker can play without "significantly" (this has yet to be defined) changing its timbre - i.e. dynamic compression.

Small 2-way speakers are always dynamically limited in the low frequency range. The short-time dynamic measurements (at 76, 86, 96, 102 dB) of @hardisj are very informative, because they introduce an objective measure for the dynamic capability.

To get an idea of the dynamic compression of the JBL 4309, I simply compared the frequency responses of the HD measurements at 86dB and 96dB.

Since these are most likely performed without changing the position of the microphone, it can be assumed that repeated measurements should result in almost identical FRs - or let's assume it was and let's further assume that the amplifier voltage was meticulously adjusted to 10dB difference .

Around 0.5dB sound pressure level difference I would define as dynamic compression, since such a level change, with a correspondingly small Q, is perceptible as a tonality change - this is simply my definition to be able to compare loudspeakers.

(I made small sound pressure level adjustments by hand, e.g. +-0.02dB)
View attachment 158607
Thus, the dynamic limit of the JBL 4309 would already be clearly exceeded in the low bass around 40Hz. The human voice range starts at about 100Hz, around 120-180Hz the 4309 reaches a deviation of 0.5dB and would thus also be dynamically at the limit in this frequency range.


In comparison, the ASR Directiva, which by this objective standard (as I said, assuming Amir's measurements are comparable) has less dynamic compression and actually only sags a bit in the frequency range above 6kHz. However, it also just reaches the limit of dynamic compression at 40Hz.
View attachment 158608

 

[More Dynamics Please]

The ability to play loud with minimal compression and distortion does seem to score more bonus points with some than with others.

 

[Chromatischism]

Maybe.
The Harman preference curves all have boosted bass. So if this is about preference then boosted bass appears essential.
I like a little boost starting from 150ish down to 40.
Have to measure in room and evaluate.
Also 100-130ish herz is often badly affected by nulls (depending on position) due to various common boundary interference so boosting it here may really be helping to slightly fill a deep dip.

I like some boost as well, but it should not be at 150 Hz, to my ears.

 

[ROOSKIE]

I like some boost as well, but it should not be at 150 Hz, to my ears.

I just start there with a gradual rise.
So at 150hrz the boost is like 0.4-0.5 db
100hrz is about 1.5-1.75db
80hrz is about 2-2.5db
60hrz is about 3.5db
40 is about 5db
20 is about 5db
This is all based off typical MMM, in room at the listening position with REW.

I do change this a little depending on a particular speaker (usually by adding a touch more boost if need be) and of course this is all done in my particular room. In another room I might like a different curve.
Who knows.
Obviously this whole curve might not work for another listener.
There is deff a point where there is to much bass as a similar curve I made with slightly more bass boost is usually a bit thick for me even though I have tried it a few times just to make sure.

 

[phoenixdogfan]

I hear you but not sure of your logic. Here, you are seeing objective measurements of this speaker with a $100,000 measurement system. Where would you go to get such data???

Getting more specific, the measurements are not the same. Outside of a few near perfect speakers, the rest have different levels of aberrations. We could sit there and guess what they mean as far as audibility or listen. I do the latter and provide that data point. In this case, you have not only me, but another reviewer, @napilopez having measured and listened with the same conclusions.

I think with many speakers, our objective clues are very revealing. Ultimately though, this is not a 100% precise science given the screwed up way audio systems are designed (we never know the tonality of music that is produced). I could paper over this and damn or praise speakers strictly on measurements. But I have chosen a balanced approach, damn the criticism I get from both camps.

I think we should be happy with 80% correlation between measurements and listening tests. The other 20% is there for us to think about. Could be that I am totally wrong in my assessment. Or that some factors in the measurements we are not appreciating (e.g. textbook perfect directivity of this speaker -- try to find that in headless panther measurements).

We could make forward progress if you could get a pair to listen to with return privileges. Do that and let us know what you think. I am quite confident this speaker will perform and perform very well.

I think this illustrates the major deficiency in the Harman preference scoring system. I remember when the Neumann KH 80 was reviewed for the third time, it's computed Harman preference score of 6.2 was nearly the same as the Revel Salon 2's score or 6.4 and my reaction was that was a nonsensical result.

I think a useful experiment here, would be to do a double blind comparison of the KH 80 with the JBL 4309 and see how subjective preferences would align compared to the computed predictions.

It's important to remember that the Harman Preference system is a scientific model, not received wisdom. At some point scientific models become less and less viable as more extensive observations turn up instances where their explanatory power is limited or nonexistent. When that happens hopefully a new model emerges which explains the new observations and subsumes the old model as a special case of the new more comprehensive one.

 

[richard12511]

something inside of me tells me the speaker wouldn't have gotten a golfing panther otherwise, considering how tiny the number of speakers that received it in the past.

Maybe so. He did still look at the measurements before posting, but at that point his subjective opinion had already formed. No doubt seeing these bad measurements before listening would have subconsciously biased what he heard(against him). He's human, so it can't be stopped.

Personally, I think this is fantastic, and why it's (IMO) so important to listen before measuring. Though I must say, all this negative feedback is one reason why he doesn't do it more often. Our ears aren't perfect, and from time to time they will get it very wrong, especially when they have no idea what they're hearing before hand. The more Amir starts listening before measuring, the more we're gonna start to see the panther ratings not quite line up with the objective measurements. That's a good thing! It's the best way to learn.

As for the negative "Harman bias" feedback(not pointed at you @abdo123), we've seen time and time again that the panther rating is based 90%+ on the results of the subjective listening test. A speaker can have great measurements yet get a 3/5(SVS), 0/5(Elac), etc. if they don't sound good to Amir's ears/brain. On the other hand, speakers that measure quite poorly(like this one) can still get a 5/5 if they do well in the subjective listen.

I was thinking this was the worst Olive score(3.7) we've ever seen get a 5/5 rating, but I had forgotten about the Revel M55XC. That Revel actually measured even worse than this, at 2.2, and still got a 5/5. Those are the two most egregious 5/5 examples I could find. Other speakers like the Genelec 8341 got a 6.8 and only got 4/5, or the SVS that got 5.7 and only received 3/5. And yes, I'm aware that Amir doesn't see the score before posting, but he does see the measurements that are very obviously much worse. Not doing this to bash Amir, either, but rather just to point out how little the objective measurements actually contribute to the overall score. The evidence shows that the score really is based almost entirely of the subjective listen.

TBH, I actually prefer it this way. By far the biggest criticism of Amir's reviews is that he judges speakers solely based of measurements, and without even listening to the speaker. It's reviews like this(and the Revel, SVS, etc.) that prove that not only does he listen to every speaker, but the final rating is based almost entirely on him doing so.

Something interesting about the two bad measuring speakers(this and the Revel) with golfing panthers is that they both have SOTA directivity. Both have a fairly poor FR, though for different reasons, but both have some of the best directivity behavior we've seen, and should take well to EQ.

 

[voodooless]

Would be interesting to know if the reasons they decided not to correct it. Possibly both budget, crossover size and audibility related, not so audible for extra complexity, size and cost of the crossover. The crossover has already quite some size and components:

Source and more interesting info about the 4309 https://www.harmanluxuryaudionews.com/2021/04/27/jbl-4309-features/

They probably did. A normal baffle step correction (using a series L/R) would still yield a bump of a few dB. You’d need an additional notch filter to get rid of it. Looking at what Amir concludes about his EQ experiment, they probably came up with the same conclusion: it’s not worth it.

 

[richard12511]

I agree with @hardisj that this would be a good speaker to put through a blind listening test! I've been looking for a speaker like this for such a purpose, though I'd want to get it used(so I could resell without huge loss).

 

[richard12511]

Nah, I decided to give it soccer panther after I saw the measurements and created the EQ and cranked up the volume. Indeed measurements here didn't make a big difference as I explained in the review.

I think he still has a point. You saw the measurements only after your subjective opinion had already been formed. The concern has never been that you're dishonest about what you hear, but rather that what you hear is different based on what you've seen(given we hear with our brains).

Here, you heard a great sound, then saw the poor measurements, then decided to still report honestly on what you heard.
The other way around, I think it's possible(likely) that your brain would trick you into hearing a worse sound than you really did(which you'd still report honestly).

Thank you for listening before hand! It's discrepancies like this that show why it's so important.

 

[richard12511]

I think this illustrates the major deficiency in the Harman preference scoring system. I remember when the Neumann KH 80 was reviewed for the third time, it's computed Harman preference score of 6.2 was nearly the same as the Revel Salon 2's score or 6.4 and my reaction was that was a nonsensical result.

I think a useful experiment here, would be to do a double blind comparison of the KH 80 with the JBL 4309 and see how subjective preferences would align compared to the computed predictions.

It's important to remember that the Harman Preference system is a scientific model, not received wisdom. At some point scientific models become less and less viable as more extensive observations turn up instances where their explanatory power is limited or nonexistent. When that happens hopefully a new model emerges which explains the new observations and subsumes the old model as a special case of the new more comprehensive one.

While I agree to some extent, this isn't just an issue with the Harman preference scoring system(Olive score). These speakers have decent(not great) FR "tilt", but they also have a shit ton of resonances. The spinorama looks quite poor with or without the Olive score judgement. At first, I was surprised by the low score, but I'm starting to agree with it more and more now based on resonances. "Lack of resonances" is one of the most important characteristics for a speaker to have, according to Toole/Olive, and this speaker does not have that characteristic.

 

[Robbo99999]

Well, the spinorama captures all these things except for the "loud" part. It captures directivity/room interaction and bandwidth.

And, to be clear, with a medium sensitivity and a rolloff starting around 100Hz, I wouldn't consider this speaker full range. Not anymore full range than other bookshelf sized speakers with a single woofer. (look at the distortion take off <100Hz as you increase volume above 86dB) Though, they did try to push it lower by extending the port tuning ... looks like they tried using a Chebyshev alignment. *shrug*

Yeah, I agree, they're about 4dB down at 50Hz from what would have been anechoic flat at that point, so imagine what the distortion would be like if they decided to boost that area by 4dB. Doesn't look very impressive to me.

 

[txbdan]

I am taking crazy pills or is that FR pretty rough looking? Since when does subjective listen testing outweigh the science on this forum? I think Amir let this speaker off easy.