Are speakers actually already a solved problem?

[anphex]

For clarity, neith our design (Sigberg Audio Manta) nor I suspect the Dutch&Dutch 8C or Geithain use the DSP directly to create the cardioid effect, it is passive / mechanical in nature. So while the speakers are active, they didn't have to be to create the cardioid effect.

EDIT: I also think it is an interesting perspective to consider active speakers to be "cheating". Care to elaborate?

I would be surprised if the cardiod optimization wasn't also coupled to the the DSP and a dedicated module running in a certain phase. But if it's rreally implemented in passive, hats off to D&D and Geithain.

Why I consider active speakers cheating: I think with a cheap DSP+Amp module you can linearize any self built two- or even three-way-speaker. You basically measure it once, apply your correction filters, and there you have +-1dB frequency response. Of course I am not considering radiation pattern and distortion here, but for the latter it comes down to cabinet weight, size and form + cone area of the woofers.

For me, if you can build something like MoFi, KEFs or even the Sierra Ascend - super well performing speakers without any active parts, that's impressive for me.

I am not shitting on active designs, they are very valid, useful and cool. I just believe it goes far more effort into passive speakers to make them perform as good as a dsp speaker.

 

[sigbergaudio]

I would be surprised if the cardiod optimization wasn't also coupled to the the DSP and a dedicated module running in a certain phase. But if it's rreally implemented in passive, hats off to D&D and Geithain.

I can't speak for them of course. Ours isn't at least.

Why I consider active speakers cheating: I think with a cheap DSP+Amp module you can linearize any self built two- or even three-way-speaker. You basically measure it once, apply your correction filters and there you have +-1dB. Of course I am not considering radiation pattern and distortion here, but for the latter it comes down to cabinet weight, size and form + cone area of the woofers.

This approach is pretty much guaranteed to not sound great.

For me, if you can build something like MoFi, KEFs or even the Sierra Ascend - super well performing speakers without any active parts, that's impressive for me.

I am not shitting on active designs, they are very valid, useful and cool. I just believe it goes far more effort into passive speakers to make them perform as good as a dsp speaker.

It's the exact same process, it's just quicker to iterate with a digital crossover. And if it's so much easier, why on earth is anyone still doing passives? Just to show they can?

 

[anphex]

I can't speak for them of course. Ours isn't at least.



This approach is pretty much guaranteed to not sound great.



It's the exact same process, it's just quicker to iterate with a digital crossover. And if it's so much easier, why on earth is anyone still doing passives? Just to show they can?

Why shouldn't it sound great? If X-over is clean and FR linear, what else is there?

And for active speakers: More parts, firmware-stack, software-stack, more personell needed for support and warranty, higher production complexity, higher production cost, higher regulatory requirements. That's the reasons to not produce active speakers I could come up quickly right now.

 

[sigbergaudio]

Why shouldn't it sound great? If X-over is clean and FR linear, what else is there?
.

And for active speakers: More parts, firmware-stack, software-stack, more personell needed for support and warranty, higher production complexity, higher production cost, higher regulatory requirements. That's the reasons to not produce active speakers I could come up quickly right now.

So active speakers are super easy, but also super complex.

 

[anphex]

So active speakers are super easy, but also super complex.

Care to elaborate? In my prior post I said clearly hat I do not consider radiation and distortion for my thought experiment. So yeah, what else is there?

And there is a small mismatch between our understanding. I come from a "I can buy a dsp+amp module (Hypex Plate amp) and use it in my DIY project". What you mean is developing, shipping and maininting a market ready product. But this wasn't the question here. I mean it all from a "I could do this at home" perspective regarding my "active is cheating" statement.

Edit: Hell, there are probably white label "hypex plate amp"-kind manufacturers that provide all the stuff I mentioned above, even the regulatory stuff, all in one package. So my point somehow also stands for commercial speaker producers.

 

[egellings]

Speakers, unlike electronics, have room interactions to contend with. So with all the different rooms that speakers can sit in, there will always be different problems to solve.

 

[phoenixdogfan]

Pretty far from a solved problem, I think. For me in order to be a solved problem, a component has to perform so well as to have its distortion and deviations from absolute perfection be inaudible AND it has to do it at a price affordable to just about anyone with a will to own it.

By that measure, DACS and AMPS pretty much fill the bill in first world markets, most second world markets, and even a great number of third world markets. A $20,000 Chord DAC will be inaudibly different form a $150 Topping. Likewise something like a Fosi V3 Mono will give any Mark Levinson amp a run for its money.

When it comes to speakers, that's an entirely different story. And there are at least three problems. First is how the sound is propagated into the room? Do we want a point source, a line source, or a panel with a wide launch area? I know of no theory of loudspeaker design which establishes the ideal way to transfer something, music, which as a live event emanates from a wide variety of sources over a wide area, into an accurate reproduction using 1-2, or let's charitably say 15 with multichannel) sources, especially given the choices those sources could take given the current state of speaker design.

Ancillary to problem one, is the question of bass, which is complex enough to be considered a separate problem. Just think or a few topics we have here: Stereo or mono? Is group delay an issue? Multiple subs needed? Is cardioid the state of the art? How about dipole bass? How much distortion is audible? L.T. .5%?, 1%?, 3%?. or greater? How low do we need to go? 40hz?, 30 hz ?, 20 hz? Infrasonic? And for what kind of playback? It might be helpful if someone could at least take a stab at whether things like the Purifi woofers which seem to bring distortions down to the level of electronics are really a material improvement and at what frequencies? It certainly would be useful know what exactly is good enough (audibly transparent?) in terms of loudspeaker distortions, and especially at what frequencies. I don't think most review outlets, including ones employing Klippel, have a handle on this.

Finally we have the issue of price and value. Do we really have a great option that gets us 90-95% or the absolute state of the art for, say, $1000/pr? Well, why not? We can get amps, dacs and streamers that will deliver most everything for l.t. $500 a unit. But speakers at l.t. $1000 a pair will have issues with bass extension or will fall apart if played much in excess of 90 db, which is well within the peak levels even casual listeners may require a speaker reach for ordinary listening. Most people even devoted audiophiles will consider a $3000 speaker purchase prohibitively expensive, and they will be absolutely correct. Such speakers achieving the above outlined levels of performance do not "solve the problem" for most audiophiles, though they do allow small capacity manufacturers to make a living by skimming the market, charging the most affluent audiophiles top dollar while selling a miniscule number of units. But those speakers only solve a handful of people's problems, and perhaps not even them if they wind up buying some gorilla coffins with staggered drivers and cabinets made of proprietary resin for the price of a 5 bedroom house in Nashville.

So no, IMHO, speakers are very far from a solved problem. Show me a cardioid design that can hit 100 db with l.t. 1 percent distortion (or just inaudible levels if it turns out 1 percent is overkill), have a uniform polar pattern down to 100 hz, be compact, and reach 40 hz +/- 1.5 db with a 50 degree or wider dispersion pattern, and not look like it was crafted by a "D+" student of woodworking, and maybe we'll be able to start talking about speakers like we talk about DACs and Amps. But that seems like a very, very long way off.

 

[kemmler3D]

I do not consider radiation and distortion for my thought experiment.

It's not a very impactful thought experiment, since (and I've never actually built speakers, but I've thought a bit about actually building some) this is where the design actually gets tricky.

Anyone can get two decent drivers in a box, and ignoring distortion and dispersion, get the crossover and overall FR flat, even passively if they are motivated to solder a bunch of parts together.

But in the same sense, anyone can build a car, just put a motor and four wheels onto a frame of some kind. Oh, you want a transmission so the wheels turn? And seats? And steering? Why didn't you say so?

Oh, but building an electric car is cheating because it has fewer parts and a third party designed the battery firmware.

 

[sigbergaudio]

Care to elaborate? In my prior post I said clearly hat I do not consider radiation and distortion for my thought experiment. So yeah, what else is there?

And there is a small mismatch between our understanding. I come from a "I can buy a dsp+amp module (Hypex Plate amp) and use it in my DIY project". What you mean is developing, shipping and maininting a market ready product. But this wasn't the question here. I mean it all from a "I could do this at home" perspective regarding my "active is cheating" statement.

Edit: Hell, there are probably white label "hypex plate amp"-kind manufacturers that provide all the stuff I mentioned above, even the regulatory stuff, all in one package. So my point somehow also stands for commercial speaker producers.

I am not sure I see the point of a thought experiment for the purpose of determining if something is easy or hard, and then not including everything it entails. I also think limiting it to your DIY project doesn't make much sense when the thread is about determining if loudspeakers are a solved problem.

Your statement and/or assumption seemed to be that all you had to do was measure the speaker, then apply enough filters to make it flat on-axis anechoically. Then it will sound great, no matter what. That's not true in any shape or form.

I've been pretty open about my development process, so you are welcome to have a look. It's not easy and it's not fast, even if it's active. I may be very inefficient and slow of course.

Sigberg Audio Manta (12" wideband cardioid active speakers) development thread

It's time for a new build/prototype thread! Thank you to everyone who followed our previous threads (Inkognito subwoofer, 10D subwoofer & SBS.1 Active speakers), all those products are now commercialized and available for purchase world-wide from our website. :D Our next project is the Sigberg...

www.audiosciencereview.com

 

[Mr. Widget]

All really fine measuring loudspeakers sound pretty similar imo, width of directivity can vary, obvious differences such as bass extension, some are engineered for a flatter target but tonally and in terms of timbre…
Keith

Loudspeakers are devices that convert the electrical signal of recorded content into an acoustic version for our listening rooms... is this last link, taking an electrical signal and making it sound identical to the original but now in our listening room solved? No.

That said, most truly well designed speakers are tending to sound very similar. Around 2010 I had the opportunity to visit Harman and listen to their double blind speaker shuffler on two separate occasions. Both times there were three loudspeakers being demonstrated in mono. Before the reveal we were not given any information as to what speakers we would be listening to. Speaker A was the then top of the line B&W 800D, speaker B was the JBL 1400 Array, and speaker C was a similarly priced Martin Logan hybrid.

The B&W and the JBL while not identical, were surprisingly similar sounding when you didn't know what you were listening to. I could have been quite happy to own either and in fact after the demo I did purchase a pair of the JBL 1400 Arrays for my home use. The Martin Logan sounded very squawky and before seeing the speakers I had assumed it was going to be a JBL horn speaker since we were at Harman.

 

[bachatero]

Why so much fuss? All good DACs sound the same because their worst measuring characteristics are below the threshold of human hearing, but the same isn't true of speakers and their distortion levels. Therefore, speakers aren't already a solved problem.

 

[anphex]

It's not a very impactful thought experiment, since (and I've never actually built speakers, but I've thought a bit about actually building some) this is where the design actually gets tricky.

Anyone can get two decent drivers in a box, and ignoring distortion and dispersion, get the crossover and overall FR flat, even passively if they are motivated to solder a bunch of parts together.

But in the same sense, anyone can build a car, just put a motor and four wheels onto a frame of some kind. Oh, you want a transmission so the wheels turn? And seats? And steering? Why didn't you say so?

Oh, but building an electric car is cheating because it has fewer parts and a third party designed the battery firmware.

You can linearize by passive crossover as precisely as with dsp? Like getting really +- 1dB? That would get you an insane network size/cost and sensivity drop, right?

All your and prior posters arguments go in a direction I can understand, but your examples are just as wonky.

A speaker with dumb dsp linearization WORKS, even if other characteristics are neglected. You car example doesn't fit here since we are talking "all parts considered" not horsepower + comfort + design + whatever. The electric car example makes even less sense.

 

[dshreter]

This seems to me akin to debating mechanical vs quartz watches. It should be possible to appreciate one (passive speakers) without diminishing the other (active speakers). Just like watches, if one has a preference for a less accurate system because you admire a certain type of complexity, that’s fine. If you prefer superior technical performance, that’s great too.

 

[Duke]

[quoting Vance Dickason] "Whether you are a loudspeaker manufacturer or a hobbyist building his own "dream" speaker design, the ultimate goal is for your speaker to sound "musical", which is another way of saying you want your speaker to sound as much as possible like the original acoustic event."

Agreed. Imo Vance Dickason chose his words wisely.

And just for the record, "sounds as much as possible like the original acoustic event" is NOT synonymous with "replicates what the recording engineer heard".

 

[krabapple]

Since the introduction of the CTA-2034 in 2015, scientific speaker design has made significant strides. The availability of more affordable measurement equipment—both hardware and software—for professionals and consumers alike has played a crucial role in this advancement.

Today, manufacturers like KEF, MoFi, Genelec, Neumann, and Kali offer speakers that are nearly perfect for their specific use cases and budgets. These speakers consistently demonstrate:

• Highly linear frequency response
• Excellent directivity
• Inaudible distortion, even at high SPL levels
• Deep bass extension relative to their size

A modern preference DBT of these a la the old Harman speaker-on-a-turntable trials, would be fascinating.

If the premise is true, the preference differences among them would be statistically insignificant.

 

[krabapple]

If we take Genelec and Neumann in comparison. They both obviously have the same goal of having a very flat response, but still (and generally speaking), fans of Genelec speakers usually don't like the sound of Neumann speakers as they often seem to find them boring and unengaging, while Neumann fans often think that Genelec speakers have an unnatural bright representation in the upper range.

If those comparisons are done sighted the conclusions aren't worth the pixels they're printed on.

We all should know that by now!

 

[krabapple]

Speakers, unlike electronics, have room interactions to contend with. So with all the different rooms that speakers can sit in, there will always be different problems to solve.

Speakers are a terrible way to solve room problems.

That is not really their proper function.

 

[krabapple]

People are defining 'solved' in various ways here, obviously.

Sean Olive's claim remains true:

The loudspeaker industry has generally accepted the science of what makes a loudspeaker sound good; there are new standards that define what is good and how to measure it, and it’s widely practiced within the industry

And to that extent, questions and problems that existed for decades, have been answered/solved.

 

[kemmler3D]

You can linearize by passive crossover as precisely as with dsp? Like getting really +- 1dB? That would get you an insane network size/cost and sensivity drop, right?

IMO almost as precisely. If we are not talking about FIR filters (which are not universal) then you only have maybe 10-20 filters to work with, and so you have a limitation that is comparable to those faced by electrical crossovers. If you look at the Ascend speakers, they are passive but you'd think they were DSP speakers.

your examples are just as wonky.

Fair enough. I was just trying to illustrate that getting a flat on-axis response without regard for dispersion or distortion ignores the actual work that goes into designing a good speaker. You are right that getting flat on-axis response is straightforward. Anyone can do it with a UMIK, REW and EQAPO if they really want to.

Analogies aside, avoiding distortion at the crossover and matching directivity at the same time is really hard, but you need to nail both of those things to make an actual good speaker.

 

[Duke]

Speakers are a terrible way to solve room problems.

That is not really their proper function.

Given that the net result at the listening position includes room interaction, whether "proper" or not, loudspeaker design that takes anticipated room interaction into account probably a good idea. Check out Sigberg Audio for an imo very well thought-out approach to taking room interaction into account.