How about a 6 channel (or 2 channel, I guess I could buy 3) DSP based programmable time delay and parametric equalizer? If we want to get fancy you could even add a calibrated microphone and some 'get in the ballpark' auto config stuff.

I'm thinking something like the ICBM, where you're solving an old problem a slightly new way, possibly creating a new niche.

It would also fit in with the 950, enhancing and also addressing some perceived flaws in that device. For instance if the 'box' was used to set a baseline delay and level the 950 settings would become 'trims', a feature some would like to see.

Just shooting from the hip - whaddya think?

Charlie

DSP based? Would you really want to add an extra A-D-A conversion into your audio loop?

------------------
Matthew J. Hill
matt@idsi.net

Want to? Nope.

But the other options are worse, unless I'm missing something. An analog device would almost certainly be less capable and have more issues, and there is no suitable means I'm aware of to stay in the digital domain.

I would, after thinking it over, go with 8 in and 8 out.

Someone already makes it:
Behringer DSP8024 Ultra Curve 24

I'm buying 3!

Charlie

http://www.behringer.com/02_products/prodindex.cfm?id=DSP8024&lang=eng

[This message has been edited by charlie (edited August 16, 2002).]

Just looked at the UltraCurve... Here's what I'd change:

20 - 20 kHz -> 10 - 22 kHz freq response
24 bit/48 kHz -> 24/192

I'd ditch the graphic eq, and add 2 more filters for the parametric eq (5 per channel).

I'd also add a subsonic filter.

Put 4 of them in a box (8 channels), and sell it for less than say $1g. I like the RTA with the included microphone!

There's no time delay in there, but I'd rather have my pre/pro doing that...

Actually I downloaded the manual and app notes. There is a time delay for each channel adjustable in 0.1 (!) ms increments.

Also, the 'graphic equalizer' is not implemented as a 31 band digital EQ, but rather the 'sliders' are used to define the 'connect the dots' response graph and a transform function is computed for the required shape. Pretty cool, plus it sells for less than $200.

It has lots of other stuff besides the EQ and delay too, but I'm not sure how useful most of the rest is for HT. Download the manual - very cool device.


I'll be buying 4 of them as soon as I have time to fool with them. The stuff you mentioned would be nice, but are not deal breakers to me.

Afterthought:

It also has a memory that allows quite a few 'themes' to be memorized and recalled by name, so if you had a 'small theater', 'big theater', 'music', etc preferences it will remember them, and it features a genuine hard relay bypass - always useful.

[This message has been edited by charlie (edited October 04, 2002).]

Yup, I saw the time delay when I fully read the info sheet.

24/48 is only CD quality though. Hopefully you don't plan on running a DVD-A/SACD signal through that.

Less than $200? Are you getting a deal somewhere? I got this from their site:




[This message has been edited by Kevin C Brown (edited October 04, 2002).]

http://www.americanmusical.com/item.asp?UID=2002081614132149&menu=&keyword=&item=BEH+DSP8024

The AD/DA is 24bit x 48kHz, quite a bit above CD level, which is 16x44. I can't hear above about 18.5 or so, and frankly I doubt you can hear above 20kHz!

For all intents and purposes I doubt any audible degradation would be worse than what is being fixed.

I would say that 192/24 is



I don't think 24/48 is that much of an improvement. Let me put it this way, my (old) Sony pre/pro, the TA-E9000ES converted all analog sources to 24/48 internally for processing. (No analog passthrough.) Even though it's a great pre/pro, its knock all along has been its "digital" sound...

I guess for the price, the Behringer is worth a try.

Well at least you can engage the bypass for DVD-Audio and evaluate whether the cure is worse than the disease. I'd rather see 96/24 but I'm not worried about it. The room effects typical in an environment will be much worse than the unwanted distortion introduced by a well done 48kHz conversion. This box has been very well reviewed and also measures very well.

24 vs. 16 bits is a pretty huge difference in range, several orders of magnitude. Sample rate is less so, 10% higher, but my experience has been that well done sampling implementations prove Nyquist correct. Obviously higher sample rates make things easier in some ways, but I'm not building a system to entertain bats. The 20Hz rolloff is more a concern, but I can live with it. It may not be a rolloff - the 0.5 down may be reflecting the rolloff at 20kHz more than 20Hz.

I think it deserves a try.

I strongly suspect the payoff is going to be much better than the added (unwanted) distortion. And, as you say, at $100 per channel I can't ignore it.

Charlie- You should post your conclusions here when you get a chance. Who knows, might spur Outlaw a little...

(Get it? "Spur" ? Hee, hee.)

(I'm personally a little curious too, because I have a high quality stereo analog parametric eq that I'm going to use for room node taming for my mains. But I haven't gotten around to hooking it up yet, because it's going to take a lot of manual measurements to do it right...)


[This message has been edited by Kevin C Brown (edited October 05, 2002).]

Just a reminder, people. Nyquist said that the minimum sample rate for reproducing a singnal is 2x the frequency of that signal. Not that you'd be able to reproduce the correct shape of the signal at that sample rate. A 22 kHz tone sampled at 44.1 kHz basically turns into a triangle wave. There are benefits of sampling higher than 44.1 kHz, even if human ears can't hear all of the frequencies that become possible at higher sampling rates.

------------------
Matthew J. Hill
matt@idsi.net

Actually that's not true. It seems like it should be, but its not. Also, a 20kHz tone, by definition, is a sine wave.

Nothing personal Matt.

EDIT:

OK - here are some ways to think of it that don't involve numbers and letters dancing together in some sort of unholy rite:

- The simple 'connect the dots' shape is the stored data, not the waveform. The information about the shape of the waveform is spread over many samples and is no longer stored in a single sample.

- The simple 'connect the dots' shape contains frequencies higher than the Nyquist frequency. Or we could say the D/A step produced these frequencies, however we want to look at it. In any case, they will be removed in the next step.

- One really cool 'high level' way of looking at it is this - the dots are not the waveform they are just data, and the only waveform that can connect the dots and not contain elements over the Nyquist frequency is the original waveform.

[This message has been edited by charlie (edited October 07, 2002).]

But what about waveforms which are not sine waves? Now I know that Fourier (I think it was Fourier?) said that they can be broken down into a sum of sine waves, but is it not still possible that a non-sine wave could be at a frequency that's within our hearing range, whereas overtones that may be present in it may not be? Couldn't our ears still tell the difference between that waveform and one that's at the same frequency but doesn't contain the overtones? If you approximate it with a sine wave I still believe that audible information will be lost.

I did not think that DACs would round off the data to produce a sine wave but now that you say it, it seems like the natural thing for them to do.

------------------
Matthew J. Hill
matt@idsi.net

The short answer is no.

Here are some things to do and think about:

The case you opened with is easy to visualize, where the wave is exactly at 1/2 the sample rate. Try a more complex case. Use a piece of graph paper and draw a 15kHz sine wave on it. Now draw vertical lines to represent the sample points at 44.1kHz. Now, either transfer those dots to a second sheet or connect them with line segments or something to show the resulting sample data more clearly. The result is a ghastly looking mess, yet the wave can be recovered.

The human ear responds to tones up to roughly 20kHz, tones being sine waves. Any tones higher are (to most people) not audible.

Obviously there will be some errors due to quantization and some problems due to flaws in the actual implementation, but it seems like the Ultracurve has a good implementation since it measures very well and has a lot of very favorable listening tests behind it as well.

I'm looking forward to trying them out.

If Outlaw was interested in this space they could implement and build a 10in/10out magic box and imbue it with capabilities far beyond what the Ultracurve does just by virtue of having all those channels in a single box, although the digital interconnect looks very flexible on the Ultracurve too.

For instance Outlaw could implement not just graphic/parametric EQ, gain and delays, but also a flexable BM system, phase vs. frequency adjustment and so on. The potential of something like this is almost unlimited, and if they could implement a digital audio out from their next higher end prepro they could even avoid that D->A->D->A issue, letting the 'box' do D->A at the last instant.

Obviously something like that would be prefered, esp. if they went 24/96 or something.

Well I got my first UltraCurve tonight, but the matching microphone is backordered until the end of the month. Grrrr.

It's a nice looking unit, anyway. And maybe I can have the menu system mastered by then....

In reading this thread I am a little confused by the DAC sampling at 24Hz issue. As a HT newbie the whole digital to analog conversion process is quite confusing. While reading a review of the new Audigy 2 soundcard over at tomshardware.com I came across a description that made sense at the time, but doesn't jive with what I (think I) read here. Could one of our myriad of knowledable sound engineer types that hang out in this forum glance at this description of the DAC process and maybe comment if what is written is accurate? I keep hearing things about differences not being audible - and sometime's I am thinking Hz refers to higher pitch or lower pitch and other times to sampling frequency... ughhhh... anyway the link is here: http://www.tomshardware.com/video/02q4/021106/index.html

On a side... the soundcard seems quite sophisticated using a Cirrus Logic CS4382 (maybe a sister of the 950 Cirrus) for decoding.

Hertz is a shorthand for 'cycles per second' and can be applied to many recurrent things.

http://www.audiovideo101.com/dictionary/hz.asp

There is an important theorem that states a digital sampling system can store all the data needed to restore an arbitrary waveform if (1) the waveform has no frequency components above 'N' and (2) it is sampled at a rate of at least N x 2.

This is where 'sample rate' becomes important.

EDIT:

I should answer the question is guess:



This is no joke - in real life it works a lot better than the picture might indicate. Also, as noted in discussion above, there is a tremendous amount of variation in quality of implementation. And for purposes of simplification toms' has omitted a few steps. See the link above for a good accurate rundown on digital recording.

[This message has been edited by charlie (edited November 14, 2002).]

I think that the sampling rate has to be higher than twice the highest frequency of interest. Just 2X won't cut it, but 2.0000001X will. A subtle (to many) distinction, but important.

I am seriously considering the purchase of one (or two) of these Audigy 2 cards to use as an audio o'scope and spectrum analyser for my audio tinkering hobby. I have never purchased a creative brand sound card because of the quality (or lack of) of the record/playback result. I have prefered the Turtle Beach soundcards because they have given me much better results, but the new Audigy 2 looks really inviting.

Paul

Paul

------------------
the 1derful1

[This message has been edited by Paul J. Stiles (edited November 14, 2002).]

Have a lookit this:

Nyquist theorem. The Nyquist theorem states that if a signal V(t) does not contain frequencies higher than fs/2 (where fs = 1/Ts), then it can be fully recovered from its sampled values V( nTs) at discrete times tn = nTs ...

http://www.digital-recordings.com/publ/pubrec.html

I want to move to theory 'cause everything seems to work there. In reality the higher than x2 the sample rate is the easier it is to implement. I suspect a x2 system might actually be impossible to build, but I'm not enough of a math genius to prove it.

There seems to be missing one very important point in this discussion of the AD/DA process.

The final step in the digital to analog conversion process always includes a low pass filter between the Nyquist frequency and the word clock frequency i.e. for a 44.1K system, between ~20Khz and 44.1Khz.

This does two very imprtant things:

It strips off the word clock (44.1K etc).

It removes ALL harmonics of the audio above the Nyquist frequency.

Thus - a 20Khz tone stripped of all of it's harmonics equals a pure SINE WAVE.

The raw output of the D/A converter before the low pass, for a frequency approaching the Nyquist frequency (20Khz) is an approximation of a square wave, not a triangle wave, due the sample and hold stage.

[This message has been edited by soundhound (edited November 15, 2002).]

Sure enough.

Actually I intentionally skipped a lot of steps in interest of clairity, at sacrifice of detail. This is what I was generalizing when I said "The simple 'connect the dots' shape contains frequencies higher than the Nyquist frequency. Or we could say the D/A step produced these frequencies, however we want to look at it. In any case, they will be removed in the next step."

Not technically complete, but I know how much I hate it when I ask a question and get an overly detailed answer to a question. It's hard to know where to draw the line.

One item that is also very often ignored or forgotten is the low pass before the sampling step.