JBL 708P Review (Professional Monitor)

[Jaimo]

The 308p is definitely the better value, imo, but that's almost always the case when comparing good cheap speakers to good expensive speakers.

There have indeed been complaints about 7 series reliability.

My personal experience has been extremely positive however. No issues after ~18 months of service. I do value the compact footprint of the 708P but if I was to do this again, I would seriously consider going for the 708i and using external EQ and amplification.

This will improve reliability and maintainability but In terms of performance, the law of diminishing returns applies strongly here.

 

[milotrain]

All the issues with regards to the 7 series reliability seems to be specific to the active monitors from the very first batch, and one of the reasons there was so much "b-stock" of the active monitors about a year ago. We have had no issues after that first batch.

 

[Edgar]

I read most of this thread. What I missed are information how to connect those speakers best to a music source. And what happens with the music signal inside the speakers. I am new to this and this forum so maybe this is obvious to everybody else.

1. The speaker has digital and analoge inputs. And it has, according to the website: "Integrated dual amplification system with internal floating-point DSP" and "Comprehensive room EQ and delay controls"
So what happens if the analoge or digital inputs are selected?
I guess with the analoge input the signal will be converted to digital, processed, and then converted to analoge and output on the drivers.
And I guess with the digital input the first AD conversion is skipped and the rest is the same? Correct?

2. What are the easiest and best ways to use these speakers at home in a stereo setup or in a home theater?
Let's say the source is a PC.
I guess it does not make sense to convert the digital output of the PC in an (expensive) DAC to analoge and input that into the speaker because then it gets (I guess) in the speaker converted to digital again.
So how should it be connected? Connect the PC USB to AES3 converter and connect that to the speakers? Or some fancy device in between?
A setup like mentioned here? https://www.audiosciencereview.com/forum/index.php?threads/jbl-705p-708p.1303/post-33768

3. What is the recommended way to setup the "Comprehensive room EQ"? Is there any recommended microphone and test setup (kit) - maybe from JBL?

4. How about subwoofers? Are they always recommended with the 708P or only i.e. with a home theater setup? Is there any specific subwoofer from JBL which is made to best integrate with these speakers?

As you clearly see I don't know what happens. Please correct me where I guess wrong and I would like to learn how to "use" these speakers.

 

[Jaimo]

I read most of this thread. What I missed are information how to connect those speakers best to a music source. And what happens with the music signal inside the speakers. I am new to this and this forum so maybe this is obvious to everybody else.

1. The speaker has digital and analoge inputs. And it has, according to the website: "Integrated dual amplification system with internal floating-point DSP" and "Comprehensive room EQ and delay controls"
So what happens if the analoge or digital inputs are selected?
I guess with the analoge input the signal will be converted to digital, processed, and then converted to analoge and output on the drivers.
And I guess with the digital input the first AD conversion is skipped and the rest is the same? Correct?

2. What are the easiest and best ways to use these speakers at home in a stereo setup or in a home theater?
Let's say the source is a PC.
I guess it does not make sense to convert the digital output of the PC in an (expensive) DAC to analoge and input that into the speaker because then it gets (I guess) in the speaker converted to digital again.
So how should it be connected? Connect the PC USB to AES3 converter and connect that to the speakers? Or some fancy device in between?
A setup like mentioned here? https://www.audiosciencereview.com/forum/index.php?threads/jbl-705p-708p.1303/post-33768

3. What is the recommended way to setup the "Comprehensive room EQ"? Is there any recommended microphone and test setup (kit) - maybe from JBL?

4. How about subwoofers? Are they always recommended with the 708P or only i.e. with a home theater setup? Is there any specific subwoofer from JBL which is made to best integrate with these speakers?

As you clearly see I don't know what happens. Please correct me where I guess wrong and I would like to learn how to "use" these speakers.

You are correct - an analog signal fed into the speaker is converted to digital (24/96) for processing in the digital domain.

There are no issues with feeding an analog signal into the speaker. Considering the many high quality low price DAC’s available, this is a safe route to start out with. You can always branch out to using the AES input and not be too heavily out of pocket.

Regarding EQ, look into the free REW PC based software option. -you will need to purchase a Microphone however.

I use my 708p’s in a stereo configuration without a subwoofer and this if fine 90% of the time. I will look into a sub at some point in the future but space and aesthetics considerations have not made this a priority.

I spent the evening listening to some of my favourite 60’s bands at high volume levels and was blown away by how clean and strain free the music comes over. This is the main difference between the 708’s vs the 308’s that I previously owned for several years.

 

[milotrain]

The JBL lsr6312sp is the sub commonly paired with it but I don’t find it necessary for the vast majority of music I listen to.

In your case I’d go AES out of the PC, it’ll be the simplest and cheapest route.

I wouldn’t bother tuning them at this stage if your room doesn’t have problems, for me they sound very good in almost all situations without tuning.

 

[BDWoody]

I am new to this and this forum so maybe this is obvious to everybody else.

Welcome! Wasn't really obvious to me either... So maybe we're both the dumbasses.

I have used mine connected both ways, and while I've never done a controlled comparison, I've never noticed any differences.

Having had a 'bad experience' with an unexpected, unattenuated near field 5.1 test signal from 3 of these and two 705P's, I would be leery of doing it straight from a computer with nothing but the slider for volume control. That was not a good time. Others may be more brave.

That said, I really like mine. Great speakers. I use them with my Integra processor's XLR outs, and haven't had any reason to think I need to do something different.

I'm a fan of subwoofers...I think there are very few speakers/rooms that couldn't benefit, but these are quite satisfying without.

 

[Kain]

What distance are distortion tests taken from on ASR? Are they all from 1 meter?

 

[echopraxia]

I’m surprised there isn’t more discussion about the fact that literally everyone who hears the 708P say they’re leagues better than the 308P Mk II, despite the CEA2034 spin measurements indicating that they should sound similar at best, and the Olive Score indicating that the (10x cheaper) 308P Mk II should actually sound better, with or without subs.

IMO this should be the final nail in the coffin of CEA2034 (or the derived Olive score) as a metric that supposedly can give a complete enough picture of a speaker to find the best speaker. Notice that whenever we are comparing two speakers whose measurements are both beyond some minimum bar of CEA2034 performance, the predictive power of CEA2034 in determining the winner seems to vanish rapidly. Therefore CEA2034 is clearly missing something hugely important.

Maybe it’s missing THD, maybe IMD, maybe it’s dynamic compression, maybe just total SPL capability, maybe it’s beam width, maybe it’s some combination of all those — who knows, but it’s exceedingly clear that something very important is missing from CEA2034, and the science here is not complete (and not just for a rare few outliers).

Either that, or everyone (including Amir) who has heard the 708P and found it to be much superior to the sound quality of the 308P either has bad hearing or is lying. I just don't believe that.

 

[Deleted member 2944]

I don't think there's any doubt about it. CEA 2034 is simply one test metric that can be used, but there have been numerous data points now that highlight relative evaluation issues.
I also don't believe the Klippel NFS is anywhere near a 100% effective testing scheme that can correctly evaluate the objective (or even subjective) performance of speaker systems. It seems an excellent tool, but there have been numerous data points and misinterpretations of its results.

In this particular case it's highlighted well because the 308P and 708P are in different leagues, performance wise. I've listened to both of them quite a bit.

Dave.

 

[ernestcarl]

CEA2034 (or the derived Olive score) as a metric that supposedly can give a complete enough picture of a speaker to find the best speaker.

I don't think anyone in the know has ever claimed that, though.

 

[respice finem]

#368: The PITA with preference ratings / scores of any kind is, an individual is not a statistic. Also, when comparing speakers (or anything else in audio) by hearing, it's absolutely necessary to do it at exactly the same SPL. Popular error with passive speakers, amp at volume "x", and an AB switch. Result: The more sensitive (and thus louder) pair almost always wins.
In this particular case, my best guess is, the higher THD makes the 308 lose in comparison. That said, I would expect most people (there we are again) to be perfectly happy with the 308 in a nearfield desktop or even small studio scenario - without the direct comparison with the 708P. Everyday listening is not comparing stuff.

 

[echopraxia]

I don't think anyone in the know has ever claimed that, though.

I cannot count the number of times that I and others have been attacked (not by you, just to be clear) as nonscientific believers in voodoo magic, when we propose the idea that a speaker with an inferior CEA2034 performance might actually be a better speaker than another one with a better CEA2034 performance.

I cannot count the number of times that members here and on other science forums have attacked me and others whenever we so much as hint at the possibility that His Divine Holiness Dr Floyd Toole might not have conclusively closed the case now and forever on the topic of speaker preference, and that maybe the science still has room to improve. Even around here, notice how whenever an anomaly is seen in CEA2034, some anecdotal comments (not scientifically proven) from Dr Toole’s book are quoted, and everyone gasps a sigh of relief that their deity has spoken on the matter and we once again have clarity with unshakable faith.

I am not here to dismiss the relevance of CEA2034 but merely to draw attention to a common trend to regard it and Floyd Toole’s surrounding commentary to have closed the case on speaker preference. There is clearly more science to be done. Yet quite a lot of people seem to want to believe that Dr Toole’s work has completely conclusively closed the case on speaker preference, and therefore react as if threatened by any claim that there may be major missing factors of sound quality not captured by CEA2034.

 

[HooStat]

Notice that whenever we are comparing two speakers whose measurements are both beyond some minimum bar of CEA2034 performance, the predictive power of CEA2034 in determining the winner seems to vanish rapidly.

I agree, and I would say that "the variance of the predicted preference increases greatly" as the reason the predictive power declines. The variability of any regression line is huge at the ends, and gets worse as you get outside the raw data used to create the regression line (e.g., above ~7). And keep in mind that the actual scores are integers so differences of a half point aren't really meaningful. And finally, keep in mind the while the scores themselves are linear and evenly spaced integers, the way people assign those scores is likely not (i.e., the difference between 4 and 5 is not the difference between 6 and 7 in terms of sound quality or preference). I think the Olive score is a measure of engineering quality that is reasonably correlated with speaker preference. There is a reason why Revel still uses blinded listening at the end of the day.

 

[Ron Texas]

It's nearly a rule around here for any discussion of speakers Genelec and Neuman will be mentioned. It's probably because of their high preference scores and a widespread belief that better numbers mean better music. As mentioned above there are some limitations to preference scores. What I suspect is speaker A having a higher preference score than speaker B means speaker A is better, but only when all other things are equal. The thing is when comparing a speaker with a 5" mid/woofer to one with an 8" all things are not equal. Likewise, with active speakers amplification appears to be a limiting factor. What's enough on the desktop isn't enough at 10 feet.

The moral of the story is don't get trapped in the numbers game and know when enough is enough.

 

[echopraxia]

I agree, and I would say that "the variance of the predicted preference increases greatly" as the reason the predictive power declines. The variability of any regression line is huge at the ends, and gets worse as you get outside the raw data used to create the regression line (e.g., above ~7). And keep in mind that the actual scores are integers so differences of a half point aren't really meaningful. And finally, keep in mind the while the scores themselves are linear and evenly spaced integers, the way people assign those scores is likely not (i.e., the difference between 4 and 5 is not the difference between 6 and 7 in terms of sound quality or preference). I think the Olive score is a measure of engineering quality that is reasonably correlated with speaker preference. There is a reason why Revel still uses blinded listening at the end of the day.

I agree that these sample size and methodology limits can cause issues, but I don’t think it’s the fundamental problem. So long as an important underlying variable remains missing from a model, no amount of improvements to sample size or test methodology will ever meaningfully reduce the prediction error.

When we try to model an N-dimensional phenomenon (irreducible below N dimensions), any model based upon M<N dimensions is theoretically doomed to suffer from this problem of “high prediction error variance”.

To make a very obvious example, consider the variable of maximum SPL capability per frequency. This metric is clearly very important to many people. You could have the biggest sample size in the world, and the best survey design ever (e.g. lacking the problems you describe with an integer preference scale etc.), and no matter what you do, you will never be able to reliably predict this factor of preference from CEA2034 data. Because this data does not measure SPL or distortion limits, any preference prediction derived from CEA2034 will forever ever doomed to have high prediction error variance when the underlying preference includes a component related to SPL capabilities.

Even if you had an infinite sample size of infinitely many listeners, speakers, rooms, test questionnaire designs, etc. etc. You could iterate and test and retest for a trillion lifetimes and so long as you neglect to measure SPL capabilities you will never, ever reduce your models error below a floor whose fundamental cause is dimensions/variables underlying the true preference which are not being measured or included in the models.

This SPL example is just to demonstrate the importance of missing a variable in a model. Whatever is going on here (and in the case of other speakers) seems to be much more than just max SPL output though. It’s probably some combination of THD, IMD and beam width, but we need more data to be confident. Too bad IMD is so hard to measure for speakers.

 

[respice finem]

...The thing is when comparing a speaker with a 5" mid/woofer to one with an 8" all things are not equal. Likewise, with active speakers amplification appears to be a limiting factor. What's enough on the desktop isn't enough at 10 feet. The moral of the story is don't get trapped in the numbers game and know when enough is enough.

This, and the probably most often forgotten fact in amateur audio: The "objectively best" speaker will only sound as good as the given listening room will allow. Many if not most rooms I've seen in people's homes are not bad in this aspect, but catastrophic. Zero symmetry, reflecting surfaces everywhere, lots of glass, and if you turn up the volume, the cupboard resonates... Even the 308 would probably be "overkill" in such rooms quality-wise, except for its mediocre power.

 

[ahofer]

I’m surprised there isn’t more discussion about the fact that literally everyone who hears the 708P say they’re leagues better than the 308P Mk II, despite the CEA2034 spin measurements indicating that they should sound similar at best, and the Olive Score indicating that the (10x cheaper) 308P Mk II should actually sound better, with or without subs.

IMO this should be the final nail in the coffin of CEA2034 (or the derived Olive score) as a metric that supposedly can give a complete enough picture of a speaker to find the best speaker. Notice that whenever we are comparing two speakers whose measurements are both beyond some minimum bar of CEA2034 performance, the predictive power of CEA2034 in determining the winner seems to vanish rapidly. Therefore CEA2034 is clearly missing something hugely important.

Maybe it’s missing THD, maybe IMD, maybe it’s dynamic compression, maybe just total SPL capability, maybe it’s beam width, maybe it’s some combination of all those — who knows, but it’s exceedingly clear that something very important is missing from CEA2034, and the science here is not complete (and not just for a rare few outliers).

Either that, or everyone (including Amir) who has heard the 708P and found it to be much superior to the sound quality of the 308P either has bad hearing or is lying. I just don't believe that.

I've been wondering about this. I have Harbeth SHL5+ in my main system, and JBL L830 in my weekend home. I like them both, but I vastly prefer the Harbeth for an immersive listening experience, even with non-bass heavy material like chamber music (the Harbeths do have a touch more LF extension). I don't think the Olive score would rank them that way. In particular, it doesn't seem like it fully describes the experience in real rooms, away from near field. Why do the Harbeth's sound so much...bigger? Compared directly to KEF LS50W, I have the same preference/observation, and the difference is vivid between my desktop Genelec 8010 w/7040 sub - the Gennies have a lovely wall of sound near field, but sound kinda small and boxy at a distance.

I do like all my speakers though. I know I'm supposed to hate the ones that aren't my top choice, but I don't. I find myself appreciating the things they do well. None of them are irritating (as I've found many speakers from Paradigm to B&W to Wilson to Magnepan to be, in different ways).

 

[echopraxia]

It's nearly a rule around here for any discussion of speakers Genelec and Neuman will be mentioned. It's probably because of their high preference scores and a widespread belief that better numbers mean better music. As mentioned above there are some limitations to preference scores. What I suspect is speaker A having a higher preference score than speaker B means speaker A is better, but only when all other things are equal. The thing is when comparing a speaker with a 5" mid/woofer to one with an 8" all things are not equal. Likewise, with active speakers amplification appears to be a limiting factor. What's enough on the desktop isn't enough at 10 feet.

The moral of the story is don't get trapped in the numbers game and know when enough is enough.

Well, CEA2034 (or any other measurement) is able to tell us when a speaker lacks any notable flaws in the measured dimensions. The reason why Genelec and Neumann are so often invoked is because they deserve it — they are among a small set of speakers which have nearly perfect CEA2034 measurements, and this at the very least tells us that these speakers do NOT suffer from a wide range of common flaws most speakers have to some degree.

This is still super useful and important. It’s likely why I have never been disappointed with any Genelec or Neumann I’ve bought without having listened to before buying — these companies do a fantastic job of optimizing both CEA2034 and THD metrics (as long as you don’t overwhelm them with extreme bass content without help from a subwoofer, which should be obvious given their design to achieve deep bass extension in a small size for people who run them to do desktop mastering work).

But clearly there are extra dimensions that matter too as we can see here, and I think it’s super interesting and important to investigate. (Now cue the set of people who will come in telling us that we mere mortals including Amir can never hope to produce any meaningfully reliable data to make any meaningful progress, and might as well just give up.)

 

[HooStat]

I agree that these sample size and methodology limits can cause issues, but I don’t think it’s the fundamental problem. So long as an important underlying variable remains missing from a model, no amount of improvements to sample size or test methodology will ever meaningfully reduce the prediction error.

Very good point, and most likely true. The tiny assumption I would add is that this other factor is completely uncorrelated with the factors already in the model. Which I think is implicit in your response already.

And, it may be that these missing factors are more important as speakers become better -- maybe distortion doesn't matter much in a speaker with issues in frequency response. Or maybe this factor can't be manipulated effectively enough without DSP. And I suppose this is why the blind testing is done at the end of the day.

 

[respice finem]

#378: Clearly, a Neumann KH 310 is better in many aspects compared with a JBL 308, my doubt is how much of this superiority will reach the buyer's ears in the average listening room. But this is not possible to compare objectively - you can send your speaker to Amir, but not your room Objective testing as ASR does it is necessary and important, but it will not be able to tell you everything about how a speaker will perform in your room, unless you have a studio-like treated room (most don't) - then you can take the results 1:1. With all desire for objectivity, we should not expect things that are physically not possible IMHO.