Software x-over function in XXHighEnd

[JohanZ]

question: if you want to try using software x-overs, do you need for eg. a second juli@ ? or......?
answer  :It will need another "sound card (Peter)
(from Gainclone heaven topic)  ".

Interesting! I'm using a Oris 200 and an Onken bas unit bi-amped. Peter could you make this software x-over function in XXHighend with 2 soundcards in the short term/near future? 

Regards,

Johan

[AUDIODIDAKT]

Haha,

That would be nice, huh
In theory its possible and already done, but as a filter behind the xx-process. (Telstar for eg.)

Now the whole point would be, when (bi-tri) amps can be build cheaper and passive x-overs that are mostly not of great quality, yes replacing them with sofware means would be the ULTIMATE goal. (also because good passive x-overs can cost +250 euros each, and no adjusting possible)
And thinking even further, how about measuring your OWN livingroom with a mic and adjust x-overs accordingly, just an example for what is coming.

But its not that easy, especially when doing it in a *proper* way, Peter can eleborate this much better.

It will take BIG resources from your pc.
As a pre-process it maybe needs a fast 6core i7, to convert to for eg. 6ch with proper speed, i dunno.
So for eg. a flac hdcd needs decoded to wav then filtered then upsampled TIMES 6 etc, you get the picture.
How to get all those channels properly routed ?, as in more soundcards or *outputs* (new/better hardware solution?)

This also overlaps a blu-ray/dtsma/dtshd part for a bit, because i convert ALL lossless multichannels to 5.1 FLAC.
When doing this, the question is now, how to get this out the pc in to several amps.

And ofcourse, the need for several NOS1's is also a problem not to forget.

my 2c

Roy

[Telstar]

It doesnt require big processing power IF the XO/EQ is done at 44.1k or 48k, but this forbids the use of Arc Prediction (i.e. with a filterless DAC the HF noise gets to your amps and speakers). and you have locked sampling rate. I'm living with this for two years now.

Second, dedicated amps are not cheaper if you aim for the best. Consider just the cost multiplied by the number of ways. But dedicated means that you can use what works best (i.e. my recipe is class D for the subs, a current amp for the woofers, the amp that you like the most for midrange and a class A 0 feedback for the tweeters).

Third, the 8channel NOS1 is needed (Peter I need some ETA please!)

If the XO is done in xxhe, all the better, because the sampling rate will be flexible and the filter is applied in the WAV files outputted, or better, the filter is at the native resolution of the file and it is upsampled together with the output. By means of drivers and xxhe settings this can be done in the most trasparent way (i.e. avoiding another layer, which is usually ASIO).
But it will take a lot of work so the wait is long.

[PeterSt]

(Peter I need some ETA please!)

I am sorry Telstar. You know that eventually I never forget your emails and all, but usually your simple questions require longer answers - and that makes me have to really sit down for it.

And you know it ... I want it myself too ...
Peter

[Nick]

I have been trying to fathom how a software cross-over filter might work. The logic I am stuck on is this;

In order to construct a digital pass filter it must surely be necessary to analyse a “N” samples at a time in order to determine frequency components of the data. I guess the size of N would be related to the lowest frequency the filter is to act on ??. What has me puzzled is that if the filter algorithm does need to analyse N samples would this lead to the filter ringing in a similar fashion to an oversampling filter ? Might this undo benefits of the faithful treatment of data that Arc Prediction and NOS achieve ?

DSPs were not really around when I did this sort of thing so I haven’t much of a clue, so sorry of this is bit of a stupid question.

Nick.

[Telstar]

There are different kind of digital filters, google "linear phase", "IIR" and "FIR".
Good reads!

[PeterSt]

Hey nick,

I sure don't want to respond to everything others don't respond to (ok, Telstar by now I just saw), but while I'm at it anyway, here is an honest answer :
I don't know.

Yet.

No signal processing in my education really, but Arc Predcition could emerge from that anyway. Dive into matters (and looking to graphs for a month - no joke) can do jobs here. And so will this (xover-ing).
All I know is that -and it is just an example- blocking DC by common (knowlegde) means will create some 5 degrees of phase shift on 50Hz already (which is huuuge). Well, *that* is not going to happen. So whatever is, it will be good, or it is not going to happen at all.

But it will, just because I set myself to it. No guarantee still ...
Peter

[Nick]

There are different kind of digital filters, google "linear phase", "IIR" and "FIR".
Good reads!

Thanks Telstar, I'v been taking a look at some of the filter theory on Wikipedia, fascinating stuff but I haven’t seen math theory like this stuff in years  .

Peter some serious maths when the time comes to look at the crossover algorithms..... respect  !

Nick.