New to this, but have been dreaming of doing this for a while... (introduction)

[sr1329]

Ever since I read about FIR crossovers I felt that while I've been anti-digital or rather anti 16/44.1 it is a format that most of my music is in and I guess I need to live with it. The first time I heard a turntable I felt digital really isn't at the level it needs to be at. I picked up an SACD player and loved it, but at the end of the day again, most of my music is in 16/44.1. With the advent of systems like Squeezebox, Sonos and Olive I see the value in embracing digital as we have it.

I suppose if I consider digital to be a "necessary evil" then why not leverage it to the fullest. I believe with my limited experience that there are a number of problems in audio playback that we can solve with digital. Starting with that most necessary evils: the crossover. I have come to believe that a passive crossover is the biggest waste of time, effort and energy. I think it alone is why a loudspeaker with $1000 in parts is worth $10,000. The design and implementation of a passive crossover is where the money goes. Then we have active crossovers which I think are a big improvement. Now that we have programs like Audiolense and digital FIR crossovers I see no reason not to jump into that action.

This is my goal: I want to build a passively cooled PC that acts my digital crossover and room correction processor. I do not want to play my music on it, I want it to be a "black box" after the setup is completed. That means I want it accept any digital signal from any playback device like a CD player, Squeezebox, Airport Express etc. and output into standard DACs.

I have no experience with the software or even with pro level sound cards, but I would like to learn from those who have more experience in those areas. However I do have a PC building background and I am excited about the many passively cooled mini-ITX silent options we have today. I suspect a dual core Atom processor won't give me room to grow so I will start with Core 2 Duo. I would use Audiolense for the software.

In an ideal world I could a MB like this: Intel DG45FC Fly Creek LGA 775 Mini-ITX Motherboard and use the HDMI port as my vehicle for the audio transmission into an HDMI DAC. Right now the only semi-acceptable option I know of is the Emotiva UMC-1 which would be the DAC/Pre for all channels. It's not an "audiophile" grade component but I'm sure its close. That would be the ideal in having just one cable, but I don't know if it is even possible to use Audiolense to control the audio coming out of that port.

Next step I suppose is to build a small passively cooled Mini-ATX machine + Lynx AES 16 (or cheaper equivalent). I would like the use 8 channels of input (with Hosa converter) for up to Four 2-channel sources and then output to 2 Lavry DA11 DACs. I wonder if this is a sound concept. Would I have issues with disparate latency? Would I even be able to use the inputs and Audiolense? I would control switching by a simple remote control, but I do not want to have to RDP into the machine other than for maintenance or re-calibration. In daily use I want it to be a maintenance free "black box" that just works (yes I know that with Windows that is a tall order).

Is anything I'm suggesting possible? I do not want play my music on this box because I have a separate Home Server that has my music and a Squeezebox system and iTunes running in parallel one using FLAC and the other ALAC. I just want silent box with 8 channels in and 8 channels out and some kind of macro based remote system that lets me switch inputs.

Can I do this?

[manisandher]

Welcome sr1329!

I totally agree that much care needs to be taken with x-overs (not just passive).

FWIW, a few years back, I 'dipped my toe' into the world of digital x-overs, but really didn't like the results I was getting. There just seemed to be something fundamentally wrong with them (certainly back then, but I suspect still today). After playing around for a while, I stuck with my analogue Pass Labs XVR1 x-overs set to 6dB/octave...

Sorry I can't be of more help.

Mani.

[sr1329]

I see, the more I look around the more I suspect this technology has not really made to prime time. There is one studio monitor that implements FIR crossovers and that is the K&H 0500. In theory they should be getting praise like no other studio monitor before it but I am not finding it. I also can't find much information on actual implementations (including all HW and SW used) and listening impressions.

I'm suddenly looking towards things like the crossoverless Zu Audio Essence. I guess I could try digital a crossover as a hobby with cheap on-board audio play with the tech. I'd like to see a major speaker manufacturer embrace this technology first.

[Telstar]

Hi Sr,

I too think that passive crossovers are a big veil to the music. So, I decided to build crossoverless speakers. The alternatives were fullrange drivers and electronic or digital XO with 2 or more drivers.
There was one last option and it is custom made drivers with natural rolloffs where the XO point would have been. I think that this cannot apply to the diyer, so I scr*pped it out.
I listened to many fullrangers and they werent truly my cup of tea - too few dynamics, too much rollof off on top, and they required huge baffles to do some bass. Moreover, I was in love with open baffles (and still am).
I did some research on drivers and software and many simulations. I ended with 12" woofers and 8" fullrange. It contraddicts the above, but I wanted something easy to do since was my first attempt, and also I didnt want to invest much. Also my room requirements imposed pretty narrow baffles (<40cm wide).

How did it go? Well, i'm still using them after one year
Software setup and trying various equalizer for XO and driver correction, and measurements were le least fun part. But after gaining confidence, changing some parameters on the fly is priceless. The 300hz asymetric XO point and the FR behaviour were more or less the same in reality. I didnt really need a long and difficult fine tune. The time spent on research paid off. Trying various filters for both XO and EQ took much longer than deciding the XO point and the depth. The software that sounded the best to me is Izotope Ozone. I havent played much to compare linear phase vs min phase filters, but both have pros and cons - the perfect filter has not been invented.
In my case i couldnt have done with a 6db XO - my FR require a much steeper highpass to keep distortion and dynamics to an acceptable (for me) level. Instead, the woofer can play clean into the khz region and I use "only" a 24db filter. For similar reasons, I also use a subsonic filter @23hz.

There are too many compromises to be made with a 2-way, especially without boxes, that i'm planning a 4-way. And i feel confident in the transparency of software digital crossovers.
There is one big drawback that i havent been able to solve and that is the sampling frequency cannot be changed on the fly by the software player.

This is not to say that passive or electronic crossovers cannot be done in a SOTA way, but they are way more expensive, ot difficult to build.
All pro digital units are too limited in the type and amount of equalization and most of them dont go over 96khz (DEQX is one of the few exceptions).
So the answer is electronic ones like Cello and Viola or passive like the Pass that Mani has.

[PeterSt]

(this doesn't respond to Telstar who posted by now, but who I had in mind for sure, knowing is desires; I also know he will be the first one to agree with all of the below -> ??)

I am sure others will totally disagree with most of my 2c :

It is true that most of the money on speaker design goes to XOver(s). So for that matter the idea might be good. That is, if you could be faster then, but that assumed. Whether the result will be better ... I wonder. There is more to it than just "connecting" two drivers, and if you only think of the impedance influences (which are not linear) you, well, could first think that's a nasty habit of analogue filtering, but you could also say that's how things just need to be formed for rolloffs etc., so that's a natural filter within itself. Either way, don't think this will be incorporated in your digital filter ... (again, which just might be a good thing).

With my ever lasting motto "a zillion things are wrong in audio", you may actually wonder what you will achieve with this. But with this I really have to say : what you will destroy elsewhere. You see, your very first degradation is around the corner already :

Right now the only semi-acceptable option I know of is the Emotiva UMC-1 which would be the DAC/Pre for all channels. It's not an "audiophile" grade component but I'm sure its close.

Do you know how much a DAC influences the sound ? I think I do, and it is the very most important piece of everything you'll have in the chain. Now, let's say a DAC which suits you (2 channel) costs 2,000. 20,000 exists too, but 2,000 is your budget and you can get something for it which is acceptable, and consistent with the rest of your chain. Now your digital xover comes into play, and now you have to choose : pay 4 x more for the DAC (assuming 8 necessary channels) = 8,000, or degrade it to a theoretical 1/8 of its quality because you don't have the budget.
When you are done with this one, you can continue with the main amps. Same story.
When you leave out the preamps (which you can), then this is it.

Stupid story, right ? well, try to shoot at it. But before you do, I have a more important one, of which I'm sure everybody in here would agree upon ... for theories :

I am working like a crazy to get some decent sound from a software player. And worse, what is good today, is better in a few months time, and apparantly there's stuff in this software area that allows us to listen to the best we ever did no matter our further chain (because it works for everybody), and this is created by software only. The VERY worst of all is : all is as bit perfect as can be, or IOW the digital data is not molested in whatever way. None.

Now think ...

This is all more fragile than electrons circuling molecules, and if I have to sneeze during a complilation of the program, SQ may be destroyed. So to speak of course, but reality is close.

I will bet you everything I own (but let's start with my wife haha) that if you put software behind it (XXHE) that just passes the audio stream through a couple of times and again does not change a single bit, the sound will be totally destroyed. Just by being there and doing nothing. Now, the next step will be "doing something" which of course is the xover processing (or whatever it is what is needed), and next you may think the sound got better from "it" (what actually ?) but you really started with destroying. So, the "it" will be the xover itself, and indeed it can be so that net you perceive this better as the analogue xover, just because you could do a better job in the digital domain.

Although the story is clear already, now ask anyone to ditch XXHighEnd and replace it with e.g. Foobar because it can do room correction. I don't think many will be able to take this psychological hurdle, which is a physical hurdle just the same (sound will be worse).

I know, there are a few people around here who use such processing (incl. AudioLense), and think they are better off net at using it. And why not. If your room is that super bad that nothing else can be done, ok. But here the tradeoff is different, because you can actually choose between analogue and digital xover filtering, the choice for digital determining the rest of your life ...

IMO there is only one way to start such a project with some sense : Let the xover processing be done in a preprocess so the original file is convoluted with it before it's played, play it with 8 channel (etc.) output, use four DACs of 4 times the value you had in mind for one 2 channel DAC, and use four main amps of 4 times the value you had in mind for one 2 channel amp.

Producing your own single driver speaker might be costing a lot less, and may sound a lot better.

Ok, these are just my ideas about this, worth 2c or less.
Peter

[manisandher]

Rightly or wrongly, I look at it like this: if we can't even get digital playback right, what chance do we have of getting digital x-over right?

How can bit-perfect players sound different? How can the same player sound different with different means of bit-perfect output? Maybe digital x-overs will get better and better as we learn more about what's actually happening in the digital domain. But make no mistake, right now we don't seem to have a clue...

But ultimately, it's about compromises right? And as Peter says, maybe the net result of using digital x-over is better for some people than using other means. But for me, it wasn't.

Mani.

PS. @Telstar, the Pass XVR1s are line-level active x-overs

[Telstar]

PS. @Telstar, the Pass XVR1s are line-level active x-overs

Yes, i was going to correct that.

[sr1329]

Hi Sr,

I too think that passive crossovers are a big veil to the music. So, I decided to build crossoverless speakers. The alternatives were fullrange drivers and electronic or digital XO with 2 or more drivers.
There was one last option and it is custom made drivers with natural rolloffs where the XO point would have been. I think that this cannot apply to the diyer, so I scr*pped it out.
I listened to many fullrangers and they werent truly my cup of tea - too few dynamics, too much rollof off on top, and they required huge baffles to do some bass. Moreover, I was in love with open baffles (and still am).
I did some research on drivers and software and many simulations. I ended with 12" woofers and 8" fullrange. It contraddicts the above, but I wanted something easy to do since was my first attempt, and also I didnt want to invest much. Also my room requirements imposed pretty narrow baffles (<40cm wide).

How did it go? Well, i'm still using them after one year
Software setup and trying various equalizer for XO and driver correction, and measurements were le least fun part. But after gaining confidence, changing some parameters on the fly is priceless. The 300hz asymetric XO point and the FR behaviour were more or less the same in reality. I didnt really need a long and difficult fine tune. The time spent on research paid off. Trying various filters for both XO and EQ took much longer than deciding the XO point and the depth. The software that sounded the best to me is Izotope Ozone. I havent played much to compare linear phase vs min phase filters, but both have pros and cons - the perfect filter has not been invented.
In my case i couldnt have done with a 6db XO - my FR require a much steeper highpass to keep distortion and dynamics to an acceptable (for me) level. Instead, the woofer can play clean into the khz region and I use "only" a 24db filter. For similar reasons, I also use a subsonic filter @23hz.

There are too many compromises to be made with a 2-way, especially without boxes, that i'm planning a 4-way. And i feel confident in the transparency of software digital crossovers.
There is one big drawback that i havent been able to solve and that is the sampling frequency cannot be changed on the fly by the software player.

This is not to say that passive or electronic crossovers cannot be done in a SOTA way, but they are way more expensive, ot difficult to build.
All pro digital units are too limited in the type and amount of equalization and most of them dont go over 96khz (DEQX is one of the few exceptions).
So the answer is electronic ones like Cello and Viola or passive like the Pass that Mani has.

Wow you have some great information! ...and experience.

My thought (if possible) is to build a PC based software driven crossover/DRC unit that accepts digital in 2 channel and outputs 6 channels with 2 XO points per channel. It seems most of the PC solutions rely on a software player on the PC itself. I'd rather just have the PC do the processing from another digital source. I'm not sure that is possible. Do you feel that something like Audiolense is limited also? I believe it can do per driver DRC/Timing control/Impluse control. In theory it would be able to tune every driver to perform such that at your listening position you get perfectly flat response and perfect timing and no phase shifts.

Am I overstating the case? I'm sure I am since this is all in theory.

I also believe that if we have allowed the "evil" of digital into our listening why not take it all the way to its logical conclusion? I think we should apply it every problem in audio right down to accelerometers on drivers or room correction. Of course there is substitute for solid design of loudspeakers or for room treatments, but I think digital can help us do that last bit of correction and do it transparently (or at least more transparently than reasonably priced analog - the Cello is out of my price range and I'm sure it would take a lot of knowledge and skill to use it to its fullest). I hope a $400 PC and $250 software and maybe a $600 Pro Audio sound card (I was thinking Lynx AES16) would solve our problems in ways that would get into the 5 figures with analog solutions. I'm just not sure if the market has yet formed around this kind of thinking. Sometimes I think at this stage the DEQX is the only sensible (but expensive) solution. I just wish it had more digital inputs. The only analog I would use is maybe a TT and perhaps an SACD player.

I will check out Izotope.

[sr1329]

(this doesn't respond to Telstar who posted by now, but who I had in mind for sure, knowing is desires; I also know he will be the first one to agree with all of the below -> ??)

I am sure others will totally disagree with most of my 2c :

It is true that most of the money on speaker design goes to XOver(s). So for that matter the idea might be good. That is, if you could be faster then, but that assumed. Whether the result will be better ... I wonder. There is more to it than just "connecting" two drivers, and if you only think of the impedance influences (which are not linear) you, well, could first think that's a nasty habit of analogue filtering, but you could also say that's how things just need to be formed for rolloffs etc., so that's a natural filter within itself. Either way, don't think this will be incorporated in your digital filter ... (again, which just might be a good thing).

With my ever lasting motto "a zillion things are wrong in audio", you may actually wonder what you will achieve with this. But with this I really have to say : what you will destroy elsewhere. You see, your very first degradation is around the corner already :

Right now the only semi-acceptable option I know of is the Emotiva UMC-1 which would be the DAC/Pre for all channels. It's not an "audiophile" grade component but I'm sure its close.

Do you know how much a DAC influences the sound ? I think I do, and it is the very most important piece of everything you'll have in the chain. Now, let's say a DAC which suits you (2 channel) costs 2,000. 20,000 exists too, but 2,000 is your budget and you can get something for it which is acceptable, and consistent with the rest of your chain. Now your digital xover comes into play, and now you have to choose : pay 4 x more for the DAC (assuming 8 necessary channels) = 8,000, or degrade it to a theoretical 1/8 of its quality because you don't have the budget.
When you are done with this one, you can continue with the main amps. Same story.
When you leave out the preamps (which you can), then this is it.

Stupid story, right ? well, try to shoot at it. But before you do, I have a more important one, of which I'm sure everybody in here would agree upon ... for theories :

I am working like a crazy to get some decent sound from a software player. And worse, what is good today, is better in a few months time, and apparently there's stuff in this software area that allows us to listen to the best we ever did no matter our further chain (because it works for everybody), and this is created by software only. The VERY worst of all is : all is as bit perfect as can be, or IOW the digital data is not molested in whatever way. None.

Now think ...

This is all more fragile than electrons circuling molecules, and if I have to sneeze during a complilation of the program, SQ may be destroyed. So to speak of course, but reality is close.

I will bet you everything I own (but let's start with my wife haha) that if you put software behind it (XXHE) that just passes the audio stream through a couple of times and again does not change a single bit, the sound will be totally destroyed. Just by being there and doing nothing. Now, the next step will be "doing something" which of course is the xover processing (or whatever it is what is needed), and next you may think the sound got better from "it" (what actually ?) but you really started with destroying. So, the "it" will be the xover itself, and indeed it can be so that net you perceive this better as the analogue xover, just because you could do a better job in the digital domain.

Although the story is clear already, now ask anyone to ditch XXHighEnd and replace it with e.g. Foobar because it can do room correction. I don't think many will be able to take this psychological hurdle, which is a physical hurdle just the same (sound will be worse).

I know, there are a few people around here who use such processing (incl. AudioLense), and think they are better off net at using it. And why not. If your room is that super bad that nothing else can be done, ok. But here the tradeoff is different, because you can actually choose between analogue and digital xover filtering, the choice for digital determining the rest of your life ...

IMO there is only one way to start such a project with some sense : Let the xover processing be done in a preprocess so the original file is convoluted with it before it's played, play it with 8 channel (etc.) output, use four DACs of 4 times the value you had in mind for one 2 channel DAC, and use four main amps of 4 times the value you had in mind for one 2 channel amp.

Producing your own single driver speaker might be costing a lot less, and may sound a lot better.

Ok, these are just my ideas about this, worth 2c or less.
Peter

I see what you're saying. This may not yet be the best use of my limited resources.

However I have of late come to believe that you do not need to spend $thousands on a DAC to get good sound. I would simply get 3 Lavry DA-11s and 3 Parasound Halo Stereo Amps. The Lavry has it's own analog remote controlled Pre-Amp. Yes I know it's likely not the best pre-amp in the world but I'm sure its acceptable.

I also believe that digital playback is fine, but transmission of digital signals along a cable is not. I also believe the first person to build a DAC with a huge buffer and clocking on board will solve most of the jitter problem. 2GB of flash memory is a trivial thing and it will hold over 3.5 hours of CD audio 16/44.1. The user will hopefully stop playback before that buffer ever becomes full. But I know nothing about digital audio design.

I'm sure our music has been subjected to all kinds of digital filtering and manipulation before we even get it. That having been said I agree that we have a long way to go to perfect our understanding of digital audio but DSP has applications in many fields and lessons learned there also apply to audio signal processing so I don't think we are that far behind. If we are then our music has already been subjected to it. So it makes sense to believe that it is okay to process it some more for your room conditions or the limitations of your drive unit(s).

But I do hear what you are saying, this technology may not be ready yet and while it may be theoretically sound its application needs some work perhaps. I could likely better spend my money on corner traps and good old fashioned listening tests and speaker positioning. It's very hard to say who's right. I have noticed a few who tried the digital way come back, and others liked what they heard. I agree it could very well be all in the head. I suppose the only to find out is to try to build a setup and compare myself (of course double blind). But I'm at a point where I just want to get a good system and I refuse to obsess over little or inaudible differences. However I do believe in long term listening and profiling my use of the system and how long I like to listen. I think the better system is revealed that way instead of concentrating hard on ABX passages.

A good part of me thinks that I should give this a few years and also secure more funding before trying to build such a comparison system. Right now I think I should keep it simple and go with what works.


Just my 2c.

[Telstar]

Wow you have some great information! ...and experience.

Thanks.

My thought (if possible) is to build a PC based software driven crossover/DRC unit that accepts digital in 2 channel and outputs 6 channels with 2 XO points per channel. It seems most of the PC solutions rely on a software player on the PC itself. I'd rather just have the PC do the processing from another digital source. I'm not sure that is possible.

It is possible. I know somebody on an italian audio forum which has build a mini-pc (atom itx based) to do just DRC. He uses a SACD player i think XO job is a bit more tricky, and at that point i think that an used DEQX (with modified dac) would be the easiest option and not too expensive either.

Do you feel that something like Audiolense is limited also? I believe it can do per driver DRC/Timing control/Impluse control. In theory it would be able to tune every driver to perform such that at your listening position you get perfectly flat response and perfect timing and no phase shifts.

I'm sorry, I havent tried Audiolense yet. It is the most user friendly software of this type. The demo does not allow to do XO, but the developers are friendly and may help you in this matter. They were on this forum a little while ago. I also have a couple more plugins to try.
But I'm waiting to do more trials to when I have a better (multichannel) DAC.
Do not forget that for XO, you need two channels per "way" of the speakers, i.e. my next project with is 4-way "hybrid" speakers will need an 8-ch DAC. I dont think anybody will need more than 8 channels unless they want to venture into HT or into some 3/4 speaker configuration, which is tempting, but fortunately i do not have the space :D

Am I overstating the case? I'm sure I am since this is all in theory.

I also believe that if we have allowed the "evil" of digital into our listening why not take it all the way to its logical conclusion? I think we should apply it every problem in audio right down to accelerometers on drivers or room correction. Of course there is substitute for solid design of loudspeakers or for room treatments, but I think digital can help us do that last bit of correction and do it transparently (or at least more transparently than reasonably priced analog - the Cello is out of my price range and I'm sure it would take a lot of knowledge and skill to use it to its fullest). I hope a $400 PC and $250 software and maybe a $600 Pro Audio sound card (I was thinking Lynx AES16) would solve our problems in ways that would get into the 5 figures with analog solutions.

To resolve most of the room problems, yes. To achieve state of the art sound from a digital source, no - that's the DAC duty and no cheap dac can do that, sorry. The good thing is that you can start with a $600 or even less soundcard and upgrade in the future, while the software foundation is already there and working

I'm just not sure if the market has yet formed around this kind of thinking. Sometimes I think at this stage the DEQX is the only sensible (but expensive) solution. I just wish it had more digital inputs. The only analog I would use is maybe a TT and perhaps an SACD player.

The market is going into the direction of diskless source, in USA faster than here in EU. Consider that most of the out of the box solutions are nothing more than little computers with linux and some software.

[Telstar]

Do you feel that something like Audiolense is limited also? I believe it can do per driver DRC/Timing control/Impluse control. In theory it would be able to tune every driver to perform such that at your listening position you get perfectly flat response and perfect timing and no phase shifts.

I want to elaborate more on this.

1) Driver equalization should NOT be performed in respect to your room and listening position.
Read some literature on Harman website and/or buy a good reference book such as D'Appolito "Measuring Loudspeakers."
It can be done of course, but if you move one speaker of just 1cm, the consequence is not nice

2) Measurements should be done in quiet big open space or in anechoic room, and in IEC/infinite baffle. Then corrections can be done passively or digitally, depending on the issues. For instance, you can fix an energy storage with a digital notch filter (albeit i did in "analog mode" with Ozone), but you cannot fix a rollercoaster impedance without a passive XO network.
That is, some serious transducer defecs cannot be successfully corrected by digital means. The solution for us is to get as faultless speakers or transducers as we can

3) Also, ROOM MODES equalization should be done with passive intervention on the listening room itself, first and foremost.

Then, what's left to the digital equalization is frequency response, group delay, crossover (with an infinite freedom that is priceless for me), and baffle/box equalization.
If you plan to build speakers as well, start with MECHANICAL equalization, i.e. baffle/boxshape and size: it'll avoid problems before they appear.

Peter is very lucky with a BIG room without real problems, so he often forgets the issues of people living in smaller and resonant rooms.

All of the above, is not to say that DRC doesnt work, but that in a proper setup room with well-built speakers its impact is not huge, and it's mostly in the bass region (<400hz) with a feeling of cleanerm tighter sound (i have heard two proper setups done with DRC and software XO besides mine which doesnt use DRC).

[Telstar]

Hm... i feel talkative this morning.

However I have of late come to believe that you do not need to spend $thousands on a DAC to get good sound. I would simply get 3 Lavry DA-11s and 3 Parasound Halo Stereo Amps.

Peter's DAC wont cost much more than those 3 lavry.
To exit with many digital channels from the PC you need an EXPENSIVE pro card. I have searched for that route and the best option was the RME AES32 which costs no less than 800€. The math is pretty simple.
If you want a low cost decent performing pc-based multichannel solution, you have to go to 8 and 12 channel external interfaces such as the Fireface, the Audiofire 8 and 12, something from Motu, Apogee (mac only). All of them can drive a power amp directly.

If you want better sound, you got to spend much more, in the 3-8k€ range. I can give you all the names, because I have researched the topic for over 2 years now.

I'm sure our music has been subjected to all kinds of digital filtering and manipulation before we even get it. That having been said I agree that we have a long way to go to perfect our understanding of digital audio but DSP has applications in many fields and lessons learned there also apply to audio signal processing so I don't think we are that far behind. If we are then our music has already been subjected to it. So it makes sense to believe that it is okay to process it some more for your room conditions or the limitations of your drive unit(s).

I believe that some of this manipulation can be corrected (to our ears) by the same means that have been used in the mastering studios
i.e. dynamic expander, properly done oversampling.
But XXHE is doing most of this aleady in the most damage-limiting way.
And some day we'll have mostly highres digital masters, which wont need any restorative manipulation at all.

I have noticed a few who tried the digital way come back, and others liked what they heard. I agree it could very well be all in the head. I suppose the only to find out is to try to build a setup and compare myself (of course double blind). But I'm at a point where I just want to get a good system and I refuse to obsess over little or inaudible differences. However I do believe in long term listening and profiling my use of the system and how long I like to listen. I think the better system is revealed that way instead of concentrating hard on ABX passages.

I had a pretty good DAC, although not recent and a VRDS transport, which i consider state of the art. I had a generic second computer with a decent soundcard (similar to the juli@, which is used by lots of people here with good results). So I did ABX before venturing into building crossoverless speakers. The difference was slightly on the plate of the spinning disc. The computer had the advantage of music storage and few clicks playback, instead of searching through a couple thousands of CDs. And it would let me play with building my own open baffle speakers (priceless). So I went that way, and as i already said, I dont regret it a single bit.

I believe that if i cannot notice a difference in A/B comparison, the change/upgrade is not worth my money. If the difference is small, then it depends (cost, usability, etc).

A good part of me thinks that I should give this a few years and also secure more funding before trying to build such a comparison system. Right now I think I should keep it simple and go with what works.

Then dont go to listen to Peter's DAC

[sr1329]

Wow you have some great information! ...and experience.

Thanks.

My thought (if possible) is to build a PC based software driven crossover/DRC unit that accepts digital in 2 channel and outputs 6 channels with 2 XO points per channel. It seems most of the PC solutions rely on a software player on the PC itself. I'd rather just have the PC do the processing from another digital source. I'm not sure that is possible.

It is possible. I know somebody on an italian audio forum which has build a mini-pc (atom itx based) to do just DRC. He uses a SACD player i think XO job is a bit more tricky, and at that point i think that an used DEQX (with modified dac) would be the easiest option and not too expensive either.

Do you feel that something like Audiolense is limited also? I believe it can do per driver DRC/Timing control/Impluse control. In theory it would be able to tune every driver to perform such that at your listening position you get perfectly flat response and perfect timing and no phase shifts.

I'm sorry, I havent tried Audiolense yet. It is the most user friendly software of this type. The demo does not allow to do XO, but the developers are friendly and may help you in this matter. They were on this forum a little while ago. I also have a couple more plugins to try.
But I'm waiting to do more trials to when I have a better (multichannel) DAC.
Do not forget that for XO, you need two channels per "way" of the speakers, i.e. my next project with is 4-way "hybrid" speakers will need an 8-ch DAC. I dont think anybody will need more than 8 channels unless they want to venture into HT or into some 3/4 speaker configuration, which is tempting, but fortunately i do not have the space :D

Am I overstating the case? I'm sure I am since this is all in theory.

I also believe that if we have allowed the "evil" of digital into our listening why not take it all the way to its logical conclusion? I think we should apply it every problem in audio right down to accelerometers on drivers or room correction. Of course there is substitute for solid design of loudspeakers or for room treatments, but I think digital can help us do that last bit of correction and do it transparently (or at least more transparently than reasonably priced analog - the Cello is out of my price range and I'm sure it would take a lot of knowledge and skill to use it to its fullest). I hope a $400 PC and $250 software and maybe a $600 Pro Audio sound card (I was thinking Lynx AES16) would solve our problems in ways that would get into the 5 figures with analog solutions.

To resolve most of the room problems, yes. To achieve state of the art sound from a digital source, no - that's the DAC duty and no cheap dac can do that, sorry. The good thing is that you can start with a $600 or even less soundcard and upgrade in the future, while the software foundation is already there and working

I'm just not sure if the market has yet formed around this kind of thinking. Sometimes I think at this stage the DEQX is the only sensible (but expensive) solution. I just wish it had more digital inputs. The only analog I would use is maybe a TT and perhaps an SACD player.

The market is going into the direction of diskless source, in USA faster than here in EU. Consider that most of the out of the box solutions are nothing more than little computers with linux and some software.

Well the Lynx AES16 is digital in and digital out, so I will need external DACs. However I wouldn't spend more than $1500 per 2 channels. I wold never use a DAC that sits in a computer. However I also don't think we really need $5000-1000 DACs for 2 channels either. The actual DAC chips today perform much better than they did even 2 years ago. Considering even high priced Accuphase products run the signal through op-amps why should we spend $15,000 on a DAC? Marketing? Audiophilia nervosa? I jumped off that wagon a long time ago. I think a competent DAC can be designed to sell for $1000-1500. The audio manufacturers want to keep the mystique up and make us believe like its voodoo to make a DAC. They want you to believe that you must throw $thousands into analog stages following the DAC chip's output. The audio parts of an iPod touch are likely no more than $20 and really the line out from one won't make you cut your ears off with a sword either. With an iMod which is a freaking capacitor added to all of $20 (if that) of audio circuitry it sounds a little more than decent. I don't think we need to spend thousands on a DAC unless one must to avoid audio neurosis, but then the money could be better spent on therapy.

I too used to believe that stuff, but $10,000 for a Preamp that measures worse than a competent $1200 Preamp? Is that some kind of $8800 tone control? I'm trying to find a middle ground, but I refuse to allow myself to be taken back into audio jewelery land. Sorry I know what I have said is like attacking the religion at its holy place, but some sense of reality is in order.

If I spend $1500 on a DAC will I get 4 TIMES the DAC if I spend $6000? I'd probably get one that looks better and maybe has some exotic traits and one that would be better to brag about to friends and to feel better about myself.  Again to each his own, but I want to approach this with some degree of sensibility.

[sr1329]

Do you feel that something like Audiolense is limited also? I believe it can do per driver DRC/Timing control/Impluse control. In theory it would be able to tune every driver to perform such that at your listening position you get perfectly flat response and perfect timing and no phase shifts.

I want to elaborate more on this.

1) Driver equalization should NOT be performed in respect to your room and listening position.
Read some literature on Harman website and/or buy a good reference book such as D'Appolito "Measuring Loudspeakers."
It can be done of course, but if you move one speaker of just 1cm, the consequence is not nice

2) Measurements should be done in quiet big open space or in anechoic room, and in IEC/infinite baffle. Then corrections can be done passively or digitally, depending on the issues. For instance, you can fix an energy storage with a digital notch filter (albeit i did in "analog mode" with Ozone), but you cannot fix a rollercoaster impedance without a passive XO network.
That is, some serious transducer defecs cannot be successfully corrected by digital means. The solution for us is to get as faultless speakers or transducers as we can

3) Also, ROOM MODES equalization should be done with passive intervention on the listening room itself, first and foremost.

Then, what's left to the digital equalization is frequency response, group delay, crossover (with an infinite freedom that is priceless for me), and baffle/box equalization.
If you plan to build speakers as well, start with MECHANICAL equalization, i.e. baffle/boxshape and size: it'll avoid problems before they appear.

Peter is very lucky with a BIG room without real problems, so he often forgets the issues of people living in smaller and resonant rooms.

All of the above, is not to say that DRC doesnt work, but that in a proper setup room with well-built speakers its impact is not huge, and it's mostly in the bass region (<400hz) with a feeling of cleanerm tighter sound (i have heard two proper setups done with DRC and software XO besides mine which doesnt use DRC).

I hear you 100%, thanks for the great info. Everything you said makes 100% sense. I need to first and foremost get a dedicated listening room and treat it properly first. Also I agree that the bass is that one area where the correction is most needed. I could see myself limiting room correction to 250HZ and below and then managing the rest of the frequencies acoustically.

Your point #2 is interesting and it makes me believe that I do not have the resources to truly perform a per driver correction. I would need to take my equipment to an anechoic chamber and fit it in an infinite baffle and measure each one. This will not be easy or even very possible at this stage for me.

Right now I am thinking about starting with decent drivers from ATC or just get the passive ATC SCM40. They have some experience in making drivers and the price is reasonable for the quality.

[Telstar]

Your point #2 is interesting and it makes me believe that I do not have the resources to truly perform a per driver correction. I would need to take my equipment to an anechoic chamber and fit it in an infinite baffle and measure each one. This will not be easy or even very possible at this stage for me.

Right now I am thinking about starting with decent drivers from ATC or just get the passive ATC SCM40. They have some experience in making drivers and the price is reasonable for the quality.

Most drivers have those measurements published in their datasheet and done by the manufacturer
Just keep in mind that sometimes are too much smoothed to cover defects. A reliable manufacturer like ATC is a safe bet, so i would trust their measurements.