JBL HDI-3600 Speaker Review

[tuga]

The confusion stems from the fact that the predicted in-room curve is a mishmash of spatial and timbral qualities. Imagine you have two speakers that are perfectly flat on axis but have very different directivity patterns. Let's say speaker A's PIR follows the ideal target curve perfectly, but speaker B is a super wide-directivity design. Speaker B's PIR will probably be tilted up a bit from A's target curve. Yet the science suggests both will sound timbrally similar, but the latter will have a wider spatial presentation.

Sorry to disagree but a diferent tilt will produce a dissimilar tonal (not timbral) balance. Timbre is more than just tone but also the envelope, and thus not wholly dependent on frequency response.
Two speakers with identical frequency response and directivity may sound tonally similar but one not reproduce the envelope as accurately.

 

[richard12511]

I do not really see them to be contradictory results, though I do think there's some merit to your comments too.

What olive is essentially saying is that the ideal in-room response of a speaker will vary depending on how much reflected sound there is, whether that comes from the directivity of the speakers or the characteristics of the room.

The confusion stems from the fact that the predicted in-room curve is a mishmash of spatial and timbral qualities. Imagine you have two speakers that are perfectly flat on axis but have very different directivity patterns. Let's say speaker A's PIR follows the ideal target curve perfectly, but speaker B is a super wide-directivity design. Speaker B's PIR will probably be tilted up a bit from A's target curve. Yet the science suggests both will sound timbrally similar, but the latter will have a wider spatial presentation.

In that quote, Olive is talking about room correction. Room correction does not affect the directivity of the speaker, it affects the speaker's response as a whole. If you have a perfectly flat on axis speaker with very wide directivity and you tilt the response down to match the "ideal" target slope, chances are you'll end up 'darkening' the sound unpleasantly. By applying room correction to match the typical target response, what you're actually doing is tilting the direct sound downward.

That said, if frequency response and directivity are equally smooth, I can imagine some speakers sound better in certain rooms than others, depending on the listener/music/preferences.

Changing EQ and changing speakers mean the exact same thing to me. They both change the sound that we hear. Yes, there is the matter of directivity, but imagine that we hold that constant. Whether you change that sound with dsp or different drivers/cabinets/ect. does not matter(imo). If it's possible for people's DSP preferences to change based on the room, then it must also be possible for peoples speaker preferences to change based on the room.

I agree in that I don't think they necessarily conflict, but do think the two points are somewhat at odds with each other, and there may be some overlap at the edges with very different dispersion designs in very different rooms.

I do think people tend to take the former study results and extrapolate them a little too absolutely. If you're trying to say that the same speakers will always be preferred regardless of room, then that's where I disagree. General statement of correlation, or high probability? sure, but absolutes are almost never true, which comes back to why I don't think these studies contradict each other.

 

[richard12511]

Why? A speaker with a given FR and directivity will naturally yield different listening position measurements in different rooms: brighter in a more reflective room, darker and more closed in a heavily treated room, and so on.

I think the point is that room curves are heavily hyped today, but aren't necessarily what our ear-brains process in the midrange and treble.

I'm not sure how your first sentence disagrees with what I said(I was asking btw).

I agree completely with your second sentence.

 

[napilopez]

Sorry to disagree but a diferent tilt will produce a dissimilar tonal (not timbral) balance. Timbre is more than just tone but also the envelope, and thus not wholly dependent on frequency response.
Two speakers with identical frequency response and directivity may sound tonally similar but one not reproduce the envelope as accurately.

I don't disagree with you

 

[napilopez]

I agree in that I don't think they necessarily conflict, but do think the two points are somewhat at odds with each other, and there may be some overlap at the edges with very different dispersion designs in very different rooms.

I do think people tend to take the former study results and extrapolate them a little too absolutely. If you're trying to say that the same speakers will always be preferred regardless of room, then that's where I disagree. General statement of correlation, or high probability? sure, but absolutes are almost never true, which comes back to why I don't think these studies contradict each other.

Fair enough - I certainly don't think there are absolutes involved here. My point was simply that there are multiple ways a speaker can arrive at a certain tilt in its in-room response and room correction doesn't account for all of this/you can't tell by just looking at the PIR or actual in-room response. In other threads, for example, people have discussed how much the slope of the PIR should be taken into account when evaluating a speaker. As Olive says, it'll depend on expected reflections.

 

[richard12511]

Fair enough - I certainly don't think there are absolutes involved here. My point was simply that there are multiple ways a speaker can arrive at a certain tilt in its in-room response and room correction doesn't account for all of this. In other threads, for example, people have discussed how much the slope of the PIR should be taken into account when evaluating a speaker. As Olive says, it'll depend on expected reflections.

Check out my edit. Took me awhile to notice that I cut out the entire first paragraph before posting.

 

[thefsb]

High five!

Right on! I have the Original 8 but otherwise the same including color. Had it a few months. Adapting to the long scale is taking a lot of practice.

 

[Ron Texas]

@GXAlan short of the expensive 4367 what speakers offer good dynamics without severe shortcomings in other areas?

 

[GXAlan]

@GXAlan short of the expensive 4367 what speakers offer good dynamics without severe shortcomings in other areas?

The Studio 5 line was also a Greg Timbers/Jerry Moro collaboration. Extra value because it is made in China and not very pretty. Supreme value when they have their sales.

I have listened to both the Studio 530 and 590. Both very good. The 590 was a bit bright in my room likely due to tweeter height. In the Stereophile review of the Array 1400, comments section, there are some great user comments about the 590. That wasn’t a speaker for me personally (it is top heavy and is not good around young children) but is a great value as long as looks aren’t important.

 

[Selah Audio]

What??? You have read D'Apolito's book and still say what you say? Keele's method is back of the envelop computation of low frequency response. It has serious issues which you cannot just wave your hand on. In the context of this speaker, it produces serious errors as Joe states his paper: Measuring Loudspeaker Low-Frequency Response by Joe D’Appolito

View attachment 61790

Klippel NFS is the generalized solution to this problem. It fully incorporates the phase in its computation of soundfield, and does not use simplification of measuring a few radiating sources and adding them with wet thumb in the air simplifications.

I suggest going back and properly reading Joe's paper and Keele's AES one. Clearly you have not understood the complexity of the problem.

I've met Mr. Klippel and he's one of the top experts in he field. I have no doubt that the NFS is capable of handling more complex designs; however, it's easy to compare the NFS results with other methods (Keele's or groundplane) for many of the speakers already tested on ASR. If you look at Erin's groundplane of the Buchardt S400 you'll notice that in the low end extension it's closer to Buchardt's NFS curve. The 1.5-2dB difference in the ASR plot needs explanation. It would be a good idea for you and Erin to test the same speaker for a future review to eliminate the possibility of a unit-to-unit variation.

With the HDI-3600's bass extension specification being way off from what you measured there's only two explanations, 1) they overstated / exaggerated the factory spec, or 2) the measurement isn't accurate. The fact that you mentioned there being a dip in the 38hz range (right where the port tuning is) points to the latter. If so, the NFS software either had a lack of reliable input data or it isn't properly extrapolating the curve from the numerous measurement points.

 

[ruinevil]

The piano black finish and design looks great in Amirm's shot, MUCH better than the chintzy looking wood veneers. I'm not convinced JBL actually designed and conceived of these to be Synthesis line speakers though, they're about double the price they should be IMO. My gut instinct tells me Samsung might be pressuring them toward higher margin product, their M.O. is reshuffling product stacks and relying on marketing/branding to push up prices without the specs/performance/materials to actually back it up.

JBL laid off their Synthesis team before Samsung bought Harman. Samsung actually hired one of the two members of the team before they bought Harman, Jerry Moro, but I think Greg Timbers retired after he was laid off. This looks like it uses M2 waveguides, which was from a separate team lead by Alex Voishvillo.

 

[jhaider]

I'm not sure how your first sentence disagrees with what I said(I was asking btw).

Because the difference in room curves stems from different rooms, not different speakers. The same speaker can yield different different room curves in optimal-sounding placement depending on the character of the room.

 

[richard12511]

Because the difference in room curves stems from different rooms, not different speakers.

That's the point. Same speaker produces different curves in different rooms.

 

[jhaider]

That's the point. Same speaker produces different curves in different rooms.

Then I'm confused.
If we accept that "room curves" for any given speaker will be different in different rooms, how does

"different speakers(that produce different curves) [would] also be preferred"

follow from

"different curves are preferred in different rooms."?

I'm not saying it can't, but I don't see any tension inherent in those two statements.

 

[richard12511]

Then I'm confused.
If we accept that "room curves" for any given speaker will be different in different rooms, how does

"different speakers(that produce different curves) [would] also be preferred"

follow from

"different curves are preferred in different rooms."?

I'm not saying it can't, but I don't see any tension inherent in those two statements.

We may have a misunderstanding about what we're talking about. I agree that the same speaker will end up with different curves in different rooms. Where I disagree is that we should be aiming for different targets based on the room. My understanding(and again, I'm somewhat asking here) is that we should be always aiming for the same target(flat anechoic). That same target may result in a different FR in every room, but that's ok, as our brains "hear through the room"( ignoring bass frequencies as those are dominated by the room). That's my understanding, at least.

For example, with this speaker, we might try to reduce that peak a bit at the crossover frequency, but we should always apply that same reduction, regardless of the room. Same would go for boosting the 10kHz dip(though someone mentioned that might not be possible). Our target should always be flat.

As for why I think aiming for different in room target curves introduces somewhat of a contradiction?

If we say that we should be targeting different curves depending on the room, then we are saying that certain curves sound better in some rooms, but not others. For me, that's equivalent to saying "certain speakers sound better in some rooms, but not others", as changing the EQ changes the speaker. The M2 with no dsp is a different speaker than the M2 with its dsp loaded.

My understanding was that the same curve(flat) is *always(I hate absolutes) preferred, though the results may look different.

 

[jhaider]

Where I disagree is that we should be aiming for different targets based on the room.

I'm not sure that's a disagreement. I don't accord much value to "room curves" in the midrange and treble. I would not aim for any in room curve in the midrange and treble, because I would not equalize midrange and treble based on listening position measurements. There are people here who jump up and down about applying signal modifications based on listening position measurements, often derived amazingly enough from single point measurements. I think that is a far fringe position.

My understanding(and again, I'm somewhat asking here) is that we should be always aiming for the same target(flat anechoic).

If by "we" you mean "product design teams for speakers," then yes that's part of it, along with smooth off axis response.

For example, with this speaker, we might try to reduce that peak a bit at the crossover frequency, but we should always apply that same reduction, regardless of the room. Same would go for boosting the 10kHz dip(though someone mentioned that might not be possible). Our target should always be flat.

Without studying the measurements in detail, I would want to know the effects on the total sound. There is room to improve most passive loudspeakers by tailored active equalization based on a family of curves, because active EQ can effect changes passive components cannot. However, applying that EQ based solely on the on axis or listening window measurements is a mistake. Generally, I would say speaker EQ (midrange and treble) should be independent of the room. I am sure there are exceptions to that though.

 

[Wombat]

I think there is also the smoothness of the curve that matters. It may be that speakers are preferred in different rooms are also ones that have superior smoothness than what is being compared.

Greg Timbers used to have a quote where he tried to capture the difference between the JBL and Revel house sound.



The live sound component I don’t have a quote for but has been summarized as follows:

Pretend you are walking through a pre-Covid shopping mall around Christmas Time. It’s loud, there is a lot of random background noise. In the distance, you hear music playing.

The SNR is poor and imaging is non existent.

BUT, you know instantly whether that music is a live band or a recording.

Greg Timbers attributed this to “dynamics.” The ability to reproduce these contrasts was a priority of his speaker designs. Develop speakers that sound more like live music, even at the expense of holographic imaging or smoothness was the classic JBL sound.

Greg Timbers goes further to explain:


And



It is worth noting that Greg Timbers uses electronic crossovers for his personal systems. But this is where JBL invested a lot in custom transducers.

Finally,



So to summarize, JBLs didn’t care about directivity errors if it improves efficiency. That’s why they have a lot of big 15” woofer designs with vanishingly low distortion.

Importantly, Greg Timbers was let go from JBL and at least in the US, Revel speakers outsell JBL Synthesis speakers wildly. The market did NOT support the classic JBL sound. However market preferences and individual preferences can be different and both be scientifically based.

A correction. The market recently has maybe not supported the JBL sound. That may be due to the shift(not necessarily preference) to toy loudspeakers for décor fashion, apartment living or even 'phones preference reasons.

This old listener will not buy a loudspeaker that can't match the transient response of old large lightweight paper cone/ large BIL drivers of old, e.g.
Altec, JBL, Tannoy, Vitavox, etal. I prefer quality recordings of 'live music' and these old speakers do it so well.

I agree with pretty much all of what Greg Timbers is quoted as saying above. It is as simple as preference.

 

[amirm]

Greg Timbers used to have a quote where he tried to capture the difference between the JBL and Revel house sound.

With heaping amount of bias without any published research behind it....

 

[Juhazi]

---
With the HDI-3600's bass extension specification being way off from what you measured there's only two explanations, 1) they overstated / exaggerated the factory spec, or 2) the measurement isn't accurate. The fact that you mentioned there being a dip in the 38hz range (right where the port tuning is) points to the latter. If so, the NFS software either had a lack of reliable input data or it isn't properly extrapolating the curve from the numerous measurement points.

Then we have the NRC anechoic measurements too, with inaccuracy below 200Hz... Their curves look like no reflex port output at all. Summed nearfield misses phase match effect and compression with high spl, like Amir said earlier.

I think that every measurement system has deficiencies and we never can overcome this. Groundplane is the best for bass and measurements at data-bass is a wonderful source! The biggest problem will always be room effect.

 

[amirm]

The 1.5-2dB difference in the ASR plot needs explanation.

It doesn't anymore. I already explained that I found the cause of this in the review:

In the process I made a breakthrough in low frequency measurements.

I just need to re-run the previous computation with higher order expansion which I will do as I get time.

There is no mystery here and certainly no justification to go back to old school measurements that have their own issues.