2. speakers
  3. 152961-11297

Considering the option of "DSP speakers"

“Go home or go big.”

I notice occasionally somebody would come in here and say something to the effect “go active” or “go DSP” or something as cryptic as “FIR”. I am more or less old fashion – big caps, big coils, and stuffs. But with all the DSP “come-ons” recently so I went lookup “what if I do DSP, what would it take, how much would it cost” … something like that. I also looked at some of the commercial self-powered speakers to see how they do it.

First of all, “DSP” is not bad as some would automatically think of it. It's the “cheap DSP” that is bad. Let's say it you were to do it, what are your options?


  1. The most simple way is to buy a “miniDSP 2x4” which is about $100 bucks. It has two analog RCA inputs, one for left and one for right channel, and four outputs, one for tweeter and one for woofer for each channel, and 2 channels as a 2way. You can download the software plug-in which is pretty simple to use for xover filter works. If you want 3way or 4way speaker, you just need to purchase two “miniDSP 2x4” modules – one module for left and one for the right speaker. The problem is you still need to purchase two 2-channel amps (or 3 or 4) that are small enough to fit inside the speaker cabinets. Most people would go with class-D amp since they are affordable and small enough to fit inside the speakers. Or if you already have an existing receiver with 7-channel amp for example, then you can use that for your amplification, albeit outside the speakers. The first problem is you still need an pre-amp and take the pre-amp output to connect to the “miniDSP” RCA input. But the REAL problem of the “miniDSP” is it would have to convert the input analog to digital by its built-in ADC to perform DSP then converting it back to analog using its built-in DAC, so it's kind of a double back and forth. For only $100, I guess the quality of the ADC/DAC probably have to be compromised. I am under no illusion to think a $100 unit will perform the same way as a $7K dCS unit. But for a low cost solution, I guess this is acceptable and if that's all you want. Personally, because of the serious compromise in sound quality, I would only use this setup for xover development.

  2. The next thing to improve is to eliminate the built-in ADC stage. Luckily, miniDSP has something called “miniDSP 2x4 HD” which is about $200, which can either take the Tos-Link digital output of your CDP or it can also take the USB output from your computer. For the Tos-Link input, I am not sure how you can control the volume control (unless your Tos-Link source has volume control), but at least you can use the USB input and control the volume from your PC or your streaming device. But even without the ADC stage, you you still have to deal with the DAC stage because the “miniDSP” still has to convert the digital back to analog, and again, being only $200 in cost, I have to assume the quality of the built-in DAC is somewhat compromised, but this time at least, you eliminate the build-in ADC stage. And same as in #1, you can either purchase a class D amp and install it inside the speaker cabinet, or using an existing 7-channel receiver. The problem with the “miniDSP 2.4 HD” is you can't buy another one to do 3way or 4way and also use the digital input, because you can't “split” the USB into two, and since the USB protocol can only allow one master, one slave at any one time, you can't “split” them. And if you use the “Tos-Link” input, I guess if your transport has two Tos-Link outputs (or your reciever has some pass-through capability” then you can purchase a second “miniDSP 2x4 HD”, then you can do 3way, or 4way speaker as long as you can provide enough amplifications.

    (BUT make sure the digital “Tos-Link” output does not have any difference in latency or delay within them. It is not something that can be sure of. If there is delay or latency between the two Tos-Link, it will definitely affect the sound since the timing of the left and right channel won't be the same.)

    Or you can just purchase a “miniDSP 2x10 HD” which has 10 outputs and has one digital input but the “2x10 HD” will cost about $500 but you only need one digital input so that would eliminate any possible latency issue. But still your digital source needs an ability for volume control to act as a pre-amp. I think if you have a PC connected to the “miniDSP” module all the time, then you can control the volume from the “miniDSP” software plug-in, but I think the point is not having any PC plug-in at all, so you can independently control the volume even without the PC.

  3. At this point, if for whatever reasons you hate the “miniDSP” corporation, you can simply purchase a set of Hypex modules, either 2way or 3way from Madisound and it has everything you need, including a built-in class-D amps and DSP processing. They are small enough that you can build them into the speaker cabinets. You still need to provide some kind of pre-amp control though, and just as with “miniDSP”, it takes analog RCA input and Tos-Link input and it also has a Tos-Link pass-through so you can connect that to the second speaker. This would eliminate the ADC stage, but you still have to deal with the built-in DAC. And of course you have to be mindful of the class D amp. Looking at the amp, it has a few tiny capacitors for power supply bypass compared to some huge caps on my separate amps and everything looks puny :-) And these things are not that cheap. A 2x4 module is about $500 and you need two so total is about $1000. That's definitely not CHEAP! (And that not including the speaker drivers).


So far, everything is simple enough. The only real reason to worry is with the built-in ADC and DAC, and therefore the sound quality could be compromised, but if you're looking at an affordable solution, then I guess all is good or good enough. Again, personally I wouldn't go with either 1, or 2, or 3 either unless I use them purely for my own speaker xover development works. Some commercial DSP speaker only use class A/B amp for the tweeter since the tweeter is sensitive to the noise of the digital amp.


4.  So what if you want to go DSP, but still want some really good quality sound? First off, at least you have to eliminate the built-in ADC/DAC all together. Use external DAC. Use external quality amplification. Luckily, miniDSP does offer a nice little “miniDSP NanoDiGi 2x8”, which does everything in digital domain. It has a Tos-Link input and has 4 digital SPDIF channel outputs (or 8 channel total). I believe miniDSP also offers a balance version so all inputs and outputs are balance, but being digital it's not that big deal. There is still a concern of the jitter of the SPDIF outputs and that can affect the sound quality on the downstream equipment. MiniDSP does offer more expensive “all-digital DSP” processing solutions but it costs quite a bit more (about $500 I think).

But still you have to provide some type of external DAC and amplification. At this point, things start to look a bit complicated (which ironically DSP promises the opposite). For a 3way speakers, you need 6DAC's, a 6-channel amplifications. There are off-the-shelf purely DAC modules that can provide multiple channels, and of course the good one will cost quite a bit of money. Or if you have purely digital amp, that is class-D amps that take the SPDIF input directly so you can actually eliminate the DAC stage all together. Just plug the SPDIF outputs of the “miniDSP” outputs directly to the amp. Again, it's the cost again. And I wouldn't be surprise if you want a better SPDIF amp, the more it will cost. Another practical thing to consider is how will these amps will fit inside your speakers or if you decide to have all the amp outside? Finding a plate SPDIF powered amp that will fit inside the cabinents can be difficult – a practical consideration. I can already imagine a bunch of cables criss-crossing!

At this point, it's no longer a straight forward “plug-and-play” but you probably have to do some research, but still it's probably not that bad.  As for sound quality, as for the performance vs cost trade off, if you just go with some run-of-the-mill external DAC, class D amp, then it why just go with option 1, or 2, or 3 and save your money.  Sure you can get the best external DAC or amplifications but then again it all comes down to cost, and if you go all the way with these, I assume it will cost a lot of money and gets complicated indeed, which goes back to the beginning, that is , is it worth it? Then why not just go with good old analog? I think you can only go so far with “plug-and-play” off-the-shelf solutions though (unless you want to turn your living room into an equipment rack). Which leads to #5.


5.  I think truly to have good DSP system is if you go “Meridian” way, that is to go “BIG” and I mean it as both figuratively and literally. Figuratively, you really need to have some real R&D, a real lab, and hiring real engineers to develop the hardware from the bottom up and everything is custom made – from the digital stage, DAC and amplification and electronic xover works with the drivers at hand. Literally, “Meridian” was referring to some of the stuffs from Meridian and I think those are the only true high-end DSP speakers. And if you can develop your own hardware in-house, then you can scale up your design and in the process, save cost but having optimal performance. But this would exclude any chance of DIY.

I've also looked at some sub-$1000 DSP speakers, and Harman Kardon “Citation Tower” for about $3000 and Elac Navis floor stander for around $2500. Some of them actually uses only class A/B amp for the tweeter, and the class D for the woofer so I guess they realize the difference in amplification quality. Anyway, the only truly high end DSP stuffs that I've have seen only come from Meridian. The others I would characterize them as “life style”, and I am sure they provide good quality sound but I wouldn't call them high end – at least not at the same level as Vandersteen or Sonus Faber, B&W, or Magico.

Most high end speakers are still analog, And I think the main reasons why there are not many out there because it costs a lot of engineering money to develop a good system and only a few big companies can afford it. You can't just plug-and-play and call it high-end. Of the off-the-shelf stuffs I've seen, I don't think they are up-to-snuff. But I also think more and more will have some type of DSP in the future and that is just inevitable.








andy2

Showing 11 responses by andy2

andy2 OP
1,563 posts
 
Having digging into the miniDSP hardware, I begin to see what I personally would see some serious short comings.  At least some of the low end miniDSP hardware, there is a limitation to the number of "taps" in the DSP section which are used to implement FIR filter.  In general the more taps are the better.  I don't have any documentation on the Hypex setup but I suspect it may have similar limitation.

For comparison, if you're using a software convolver (a bit technical), the number of tabs is about 16K.  The miniDSP on the other hand only has 1K of taps! and that will definitely affect the fidelity of the high frequencies.  And from listening to a few room correction setup, it's the high frequencies that are most difficult to get right.

Also some of the room correction filters used by miniDSP are of IIR type (FIR is preferred) so that's another thing that will adversely affect the high frequencies.  MiniDSP does offer a hardware module that will get you up to 16K of "tabs" but it gets more and more complicated and at that point you wonder it it's all worth it.

That's not to say DSP is inferior.  It's just mean  you are limited by off-the-shelf solution as I said in my previous post, so you have to develop your own hardware solution to get optimal performance which at the end will mean it does cost quite a bit to get good DSP.  I would envision that if you develop your own hardware, everything will be on a single PCB board so it won't be like plug and play with multitude of boxes and modules.

Another problem with using miniDSP is that there is no integrated software solution for simulation.  So whenever you change your miniDSP filter, you can't simulate to see the affect.  So you end up having to measure each time you make any changes to your miniDSP filter.
andy2 OP
1,563 posts
 
Also as for the Hypex plate amp mentioned above, I am not sure if the built-in digital amp is "SPDIF amp", if so then it’s possible the Hypex module does not have to go "digital to analog" conversion which is definitely a plus. Their documentation is a bit sketchy so I am not sure, but of what I could read from the website, there is some implication that their digital class-D is all digital, that is no "digital to analog" conversion is needed.  So if you go with Hypex, then the ADC/DAC conversion is completely eliminated.  I guess that is one advantage over  miniDSP.  
andy2 OP
1,563 posts
 
A "DSP" manifesto?  An audiophile brave new world? ... Can you apply it to my bedroom?
https://www.dirac.com/dirac-blog/room-correction-vs-room-eq

Looking forward, Dirac’s exploring ways to apply its leading room correction technology to more dynamic and complex spaces, such as kitchens, where there’re more reflection points and more widely varied seating positions than a home theater. When mass-market consumer products begin integrating room correction, an entire home could be blanketed with acoustically perfect sound. We also hope to see room correction solutions for tube amplifiers in the next few years, which will offer audiophiles complete optimization of their high-end two-channel music systems.



andy2 OP
1,563 posts
 
It seems like Dynaudio also offers some pretty high end "active" options. Their most expensive is the Focus XD 60 which is $10K even. Why couldn’t they make them $7K even? I might throw in a $1K tip lols.

$10K is some serious money for a pair of speakers. But then consider a pair of Wilson for $200K. And it’s not even active.
andy2 OP
1,563 posts
 
Colonel Kurtz "You don't like my methods?"
Captain Willard "I don't see any method."

Since we're on the topic of "DSP", with regards to some of the methods, it seems like some of them try to correct the entire frequency spectrum, which maybe a little overdone.  I've listened to a few and although it is a bit interesting and it does make the sound a little "clearer" with "tighter bass", but something about the treble that doesn't sound right.  You could see right away the sound is somewhat processed not not quite "real" - makes for some nice party music I guess..

I think room correction should be done in the context of the bass frequencies.  Not the mid range. And certain not the treble.

Someone mentioned Vandersteen.  I think that may be preferable.  They only try to "room correct" the bass which usually is most of what people want.  

andy2 OP
1,563 posts
 
An interesting excerpt from an interesting article regarding to "room correction:

An example will show this point more clearly. Consider a loudspeaker standing in a room. Mr A measures impulse responses in a certain listening volume and finds to his dismay that the magnitude response has a substantial broad dip at some rather low frequency, say 300 Hz, in all positions. He calibrates a peak filter and fills up the hole in the magnitude response, which is then confirmed by measurements. Enter Mr B. Mr B is a musician and he listens to the equalized system. “It sounds horrible! What have you done to the system!? It sounds all swollen and strange!” Mr A becomes nervous, as Mr B is an important customer, and calls his trusted friend Mr C. Mr C answers: “Ah, yes of course. The dip was really due to reflections. You should never
Dirac Research AB 5
boost any dip, because they are typically due to reflections.” So Mr A removes his equalizer filter and lets Mr B listen again. Mr B, however, is still not happy. “It is better, but it’s not good. There is something hollow about the sound.” At this time Mrs D enters the conversation. She’s been listening, sitting quietly in a corner of the room, and says: “Mr A was wrong because he forgot about the time domain. Looking only at the magnitude of the Fourier transform and interpreting it as strongly related to our concept of frequencies, he thought that he could boost that region and obtain better sound. The problem is that he uses minimum-phase filters and consequently adds energy at that frequency early in time. But if we only look at the direct wave there is no hole to be filled in the frequency response. The hole never exists if we look at a short window at any time.” Mr B frowns: “So Mr C was right to say that we cannot do anything about it. But if that’s the case, why do I still hear a strange sounding oboe on my recording?” Mrs D looks sternly at him: ”Mr C was wrong too. The problem is due to the time domain properties; the reflection causes the problem and it can only be corrected for by a time-domain approach. If we design a filter that reduces the reflection, you will end up with the interesting result that the hole will be gone and the oboe will sound more natural.” “But,” Mrs D adds, “don’t take this example as evidence that you can always correct dips this way! In this case it was possible, because all positions experienced the same problem.”

andy2 OP
1,563 posts
 
One of the main motivation is that one can implement any forms of filtering - your imagination is the limit.  And of course, if you want to implement "time coherent design", using passive filter may be very difficult, but with "DSP", it's much easier.  That is your speaker can have 0 phase shift, proper step response and so on.  

As it happens that if you go with miniDSP solution, there is a program called "rePhase" that is run in Phython, that can be used in conjunction with miniDSP software, that will implement the xover that will give you "perfect 0 phase shift, time coherent design".  The only constraint is you speaker has to use "LR" filter such as LR2 or LR4.

Or if you're pretty good with DSP, you can implement yourself using the same strategy as "Bang and Olufsen" uni-phase approach which doesn't seem that complicated, that is as long as you have a background in filter theory and DSP.
Here is a link to the article the "B&O" approach.
https://www.tonmeister.ca/wordpress/2015/10/29/bo-tech-uni-phase-loudspeakers/

Assuming you have a three way speaker - woofer, mid, and tweeter.  Each will have it's own filter - and therefore will have its own filter transfer function. 

Tweeter transfer function: FS1
Mid transfer function: FS2
Woofer transfer function: FS3

So you have the total speaker transfer as : FS1 + FS2 + FS3 = FStotal.

Now most speaker with inverted polarity and so on, the total speaker response, FStotal, will have some phase shift - be it 180 degree or 360 degree and so on.  So in order for a speaker to have 0 phase shift, time-coherent response, FS1 + FS2 + FS3 will have to be equal to "1" that is:

FS1 + FS2 + FS3 = 1

Here is how "Bang and Olufsen " approached it:
First they implement the speaker as a two way - with just the woofer and tweeter - in this case, initially you have only FS1 + FS3 = FStotal.  But FStotal is not yet "time coherent" and FStotal is not yet equal to one "1". 

At this point, what "B&O" did was using FS2 (the midrange) as a sort of "tuning", to turn FStotal to "1".

In term of equation, here how it goes.

FS1 + FS2  + FS3 = 1   (This equation is made to be "1".  We will solve for FS2)

FS2 = 1 - FS1 - FS3.  (Here is what FS2 has to be, to get FStotal to "1")

So as long as you implement FS2 according to the equation above, that is:  FS2 = 1 - FS3 - FS1, your speake will have perfect time-cohernet response.

Of course, you probably don't have a three way design, but similar approach will get you time-coherent, if a little more complicated.


andy2 OP
1,563 posts
 
https://vanatoo.com/shop/speakers/transparent-one-encore/
Does it come with a glass of wine :-)

It's hard to imagine how they could manufacture something like that for only $500.  Probably done in China with people making about $300 dollar for the entire month.  And they probably would just dump whatever waste products right into the sewage system.

Most of us would probably feel guilty seeing how all the cheap Walmart stuffs being manufactured in China.  Most would cry in horror at the working conditions.  People there probably have a hard time affording some basic stuffs such as having clean food and water.  
andy2 OP
1,563 posts
 
Along the line of DSP, I stopped by a dealer a couple of days ago, and he got some top-end room correction software and He gave me a demo but although in some sense it does sound "good", but again the treble just don’t sound right, it’s like the treble sort of came out of nowhere, like it was recorded from another recording and mixed together afterward. After listening to a few room-correction setup, I think the holy grail is the treble, which can really spoil a good listening, and I don’t mean the setup sounds "harsh", in fact the treble sounds fairly smooth, but the DSP did something to it that just doesn’t sound right - it’s something I guess you have to listen. The bass is OK, and I didn’t mind as much although it lacks a bit of warmth but then it’s not something that it would bother me.

Is the cure worse than the ailment?  I then went back to my home, and listened with all the room reflections, resonance and all that.  Imperfect, yes, but 100% natural.  
andy2 OP
1,563 posts
 
Most people have never been to China. Most people have never set foot in a Chinese factory let alone many Chinese factories. Those people write things such as this which are founded in ignorance. That is not to say things are perfect there, obviously, but this statement is pure ignorance.
Although I have not been to everywhere and every parts in China, my company and my family have been there.  And the experiences have told me it's not all it is supposed to be.  There are a lot of hypes with China and I am sure we all know where the hypes originating from (hint ... not coming from China).  

I don't mean China will blunder into chaos, but you could see how "unstable" politics can get from "stable" into something completely undesirable if using Hong Kong as an example.  

Also from the economics point of view, in order to produce extremely cheap products (sold at Walmart and so on ...) somebody has to suffer.  There is no really free lunch.  If I get a free lunch, somebody else has to pay and it's a matter of mathematics.  Sure China has done a lot better now comparing to say 30 years ago, but still, there are still a lot of problems there and most people are rather poor.  

andy2 OP
1,563 posts
 

I’ve voiced my opinions on some of the issues with room correction if it tries to correct the entire freq. band, bass to 20KHz, specifically to the treatment of the treble audio range.

Another philosophical problem is that, let’s say you purchase a pre-amp with room correction capability, and of course, I assume that this pre-amp should work with any system, any amps, and of course any speakers.

Now if this pre-amp will use the system (speakers, amps ...) to measure the response of the room, there is no way it can separate the anomalies that come from the room or from the system. For example, if the speakers have a freq. bump at 7KHz, or a slight dip at 3KHz, the correction software will have no way of knowing if the causes come from the system or the room, unless the pre-amp has a separate sound generator (its own speaker), so that it could separate out the room anomalies or the system anomalies. But still there are other potential issues:
1. on the bass freq, the pre-amp sound generator won’t have the same room sound pressure signature as the system speakers, and of course the pre-amp speaker may not generate enough sound pressure to excite the room different resonant nodes.
2. on the upper freq, the system dispersion pattern won’t be the same as the pre-amp speaker, and the fact that the system speaker will have separate tweeter, mid, bass drivers ..., so you’re still left with the fact using your own pre-amp speaker is not going to help, at worse case, it may introduce extra variables that make things worse.

So, I think you’re still better off with using the system speakers to measure the room anomalies, and that means you have to correct the entire freq. band, bass to 20KHz, which goes back to the original problem. You can’t separate out the anomalies of the system and the room. And of course, for example, the speaker may have a slight dip at 3KHz or some bumps at some other freq but could be done intentionally for a reason and you definitely don’t want to "correct it out" against the intent of the speaker designer. Maybe you can have some fancy algorithm to isolate the room anomalies but from what I’ve read so far, not too many good options. I don’t know ...

Or the other option is only correct for the lower freq and do not touch the upper freq. It’s like the lessor of two evils.

And just to be clear, what I said above is only applicable for home audio system. With a professional setting such as a movie theater, then it’s a whole new ball game.
More to discover