One more quick thought. Last night I picked up the record by Julie Roberts (country artist). This record was recorded and mixed by Gary Paczosa, and it sounds absolutely stunning. This guy is SO good. Somewhere in the back of my mind, I recall that GP is a Nuendo user (am I right about this)? Whatever format he's using, he really makes it sound amazing...warm, clear, crisp, etc. It really, really sounds good. I'd love to know which format it was. BRAD, do you know? Anyway, I would encourage every one reading this post who is an AE at any level to check this CD out for its sonics.

Mr. Truth

I NEVER said that.

We have something in common.

Nika.

Au contraire, it still noise... similar to a floating point representation, the noise floor is now floating, and it depends on the density of the neighboring pulses, but that's only half of the story. The other half comes from the fact that the bitstream produced by the converter is not a representation of an instantaneous sample value but a representation of the signal slope.

And this is where we go back to the question "What is a differential bitstream?"

A differential bitstream represents the differential between two points on the signal (in DSD case, or to be precise in sigma-delta converters, the slope) and it signals whether the difference between two consecutive samples increases or decreases. In this case if the sample rate is high enough (And 2.8 MHz qualifies as high enough), this approach yields a very accurate following of the signal slope. Pulse Density Modulation (As per the raw output of a sigma-delta converter) is one type of differential bitstream, for example, all forms of FSK modulation produce differential bitstreams (Opposite case PSK forces you to do differential encoding before modulating).

Now, when the converter saturates it's only affecting the slew rate of the signal as opposed to the instantaneous sample value, so although the noise floor is higher and uncorrelated this is doesn't degrade the audible portion of the reconstructed signal, the quantization noise is definitely beyond the stop band of the DSD DAC LPF at all times and for all samples, and is also below the timing sensitivity threshold of the ear.

Reefman paper (Where Merging got their graphs from) shows the DSD response as it is, full of broadband noise, noise which as I said, is beyond the DAC LPF.

Food for thought, which one is more sensitive to jitter? DSD or PCM?


Yes, but that doesn't actually mean that it's broken in itself as a principle, it means that the quantization error can't be decoupled, only averaged over time... statistically though, the error average should decrease over time. That whole issue of randomizing behaviour to force linearity is a double edge sword, sure your noise floor is independent which assists during processing, on the other hand, if the signal exhibits a non-linear behavior you are screwed.

BTW careful with Betamax vs. VHS comparisons, Beta was a better format from a technical standpoint (Chroma, Luminance, Contrast, tape sensitivity and so on)... VHS won the war because Beta was delayed in court.

[QUOTE]Originally posted by Zoesch
Au contraire, it still noise...

If the noisefloor is correlated to the signal then it has distortion components in it. For the sake of conversation we do indeed call it noise, and to some degree it has noise in it because it has the random behavior of the DSM at its root, but it is actually not random enough to call true noise. This is where the problem enters - because it is distortion, certain signals can cause the distortion to reveal itself. In other situations it does indeed behave like noise.

similar to a floating point representation, the noise floor is now floating, and it depends on the density of the neighboring pulses, but that's only half of the story.

I agree that it is similar to floating point. Fortunately we get away with it in floating point because the distortion created is lower than the noisefloor of the audio itself. So we can still refer to it as a "noisefloor" because there is indeed noise at a higher level than the distortion. For DSD this is not always, or inherently the case.

The other half comes from the fact that the bitstream produced by the converter is not a representation of an instantaneous sample value but a representation of the signal slope.

Signal slope? What?

A differential bitstream represents the differential between two points on the signal (in DSD case, or to be precise in sigma-delta converters, the slope) and it signals whether the difference between two consecutive samples increases or decreases.

No. That would be a delta modulator, not a delta sigma modulator. If it really did this then what would be the representational difference between a sine wave and a triangle wave of the same frequency?

In this case if the sample rate is high enough (And 2.8 MHz qualifies as high enough), this approach yields a very accurate following of the signal slope.

No, you're still stuck on delta modulators. These are delta sigma modulators, and they are a straight 1 bit PCM system with noiseshaping (the noise-shaping being the delta-sigma part of the delta sigma modulator - subtract error from quantization step and feedback to next sample then add it - delta then sigma.) This is very different than a "differential bitstream."

Now, when the converter saturates it's only affecting the slew rate of the signal as opposed to the instantaneous sample value,

No. That would be true with a delta modulator, but this is delta-sigma, and it is not differential like you are describing. Slew rate is totally unrelated to the encoding method of this system.

Reefman paper (Where Merging got their graphs from) shows the DSD response as it is, full of broadband noise, noise which as I said, is beyond the DAC LPF.

But the paper doesn't show the distortion that can result as a byproduct of very low frequency, steady state signals at low amplitude. The feedback circuit starts looping constant data so that there is complete correlation between the quantization error and the signal, resulting in audible distortion characteristics. Best way to test this is with a little bit of DC going into the converters at low level - the repeating cycles in the DSM's feedback loop present it as low level birdies that stand out above the noisefloor. That's distortion, and it's in the audible range.

Food for thought, which one is more sensitive to jitter? DSD or PCM?

In a delta modulator I'd say the PCM system. But this is a delta-sigma modulator and they are equally affected by jitter.

That whole issue of randomizing behaviour to force linearity is a double edge sword, sure your noise floor is independent which assists during processing, on the other hand, if the signal exhibits a non-linear behavior you are screwed.

I didn't understand this last part.

BTW careful with Betamax vs. VHS comparisons, Beta was a better format from a technical standpoint (Chroma, Luminance, Contrast, tape sensitivity and so on)... VHS won the war because Beta was delayed in court.

Yes, exactly my point. Just because a product wins on the market - or is introduced on the market - doesn't mean that the product is inherently superior. That was the point of my beta vs. VHS analogy.

Nika.

[QUOTE]Originally posted by Nika Aldrich

Quote:

Originally posted by Zoesch
Au contraire, it still noise...

If the noisefloor is correlated to the signal then it has distortion components in it. For the sake of conversation we do indeed call it noise, and to some degree it has noise in it because it has the random behavior of the DSM at its root, but it is actually not random enough to call true noise. This is where the problem enters - because it is distortion, certain signals can cause the distortion to reveal itself. In other situations it does indeed behave like noise.
Nika.

If you call it a noise floor then there is noise.....if there is no noise in the signal then there is no noise floor.....

How about a "distortion floor"?

Yes, we do call it that. If the quantization errors are decorrelated from the signal enough that the behavior at the "floor" is random then we call it a noisefloor. With DSD it is possible to have the quantization errors correlated to the signal enough that the errors produced exceed the random errors, so the floor becomes a composition of distortion and the term "distortion floor" is appropriate.

Modern PCM converters (say, last 4 years or so) get around this problem in a way that is simply impossible with only 1 bit to play with. Prior to that, and back to about 1989, PCM converters suffered from the same problems.

Nika.

you go, boy! send 'em back to school.

oh, and zoesch? it seems you've been squoesched.

Loudist, of course some of this pisses me off. Just like I hate the high interest rate on credit cards and I think lawyers make too much money. But at the same time, this is America and I don't really hate a company for trying to take over a market. Do I think it's good for our industry? Hell, no.

On the other hand, I am able to separate my feelings about the business side from the product called Pro Tools, which overall I think is an absolutely kick ass product. Is it the best? Well, I don't know because, as stated above, I haven't had the opportunity to work with many of the others. But in general, I think the entire DAW thing is a great one...coming for the era of 2" tape, I thank god every day when I'm able to sit at my computer and do what we can these days.

So, basically, I guess I'm agreeing that the monopolization of our industry is a regrettable situation, but overall, business considerations asides, I still love Pro Tools. And I'd probably love some of the competitors just as much, if not more.

Mr. Truth

nat, i don't doubt that loudist posted that. i'll make a long story short: he's on my ignore list. several others have told me some other things he's posted about me. one of these, whom loudist believes is utterly loyal to him, told me that he thinks this is loudist's way of asking me for a date.

i love your idea about the avatars dating, but i fear that when my avatar picks his up for their date, loudist's avatar will faint dead away as soon as he opens the door.

what the FUKK?

this has turned into general hospital.



(gasp)
(Fred talking to Mary Do you mean to tell that you... (sob) SLEPT with Loudist?

(Mary looking forlorn Please.. please. I can't hear another word. I feel as if i shall faint...

But Mary! I need to know if you had... relations with him...

No. No I didn't. (looking away.. eyes filling with tears...) Oh, I can't take it anymore.... YES! YES I DID!! There! Does that make you happy?!!

(Fred, at this point goes back to his Digidesign system... a cheap 001. cheap, but something he paid waaayyy to much for) I KNEW IT... I can't stand it anymore. I've gotta chuck this piece of shit.

(Mary looking quite perplexed) What? Darling, I still love you...

(Fred begins unscrewing the 001 out of his highly overpriced rack unit purchased from Sweetwater and grumbles softly to himself..)

(Mary staring in disbelief) Darling? Darling, i still love you...

(Fred at this point throws the 001 out the window and proceeds to call his best bud, Eric. Eric from ASCAP, that is.)

(Mary) Darling? Darling are you listening to me?

(Fred) Hey buddy, it's Fred.

(pause) yeah, I'm ok.

(pause) uh-huh...

(longer pause) well, if she tries to steal your lunch money again, just tell her 'no'..

anyway, I wanted to ask you something. I picked up this box the other day cuz this dude at Sweetwater told me to. But I can't figure out what its for...

huh? well, it says D.I. on it, but I can't seem to figure it out...

what?

you don't know?

yeah...

I guess I can always call Sweetwater then...

ok, buddy.

(click)

I wish everyone would shut up about Pro Tools and just go away and be innovative.

I use PT and have always resented their monopoly
on hardware. They have not displayed the same
ability to design hardware as they have with their
innovative software. That said, there are a few choices. Apogee (the new X series is on order for me) and Prism. Both are better than Digi's hardware.
best,
chap

umm, maybe I'm replying to the wrong thread?

I'm not taking sides or deliberately trying to fuel the fire on this deal, but here's even more charts, graphs, and commentary ... if it still matters...

http://www.smr-home-theatre.org/surr.../page_07.shtml
http://www.smr-home-theatre.org/surr.../page_08.shtml

[QUOTE]Originally posted by loudist

Upper sampleing rates of PCM (96k), although on paper, should sound a lot better overall,

Depends on which paper, obviously. I'm not surprised you had the experience you did.

and if you know anything about me, I will be brutally honest about what I think


...I can't imagine.

Cheers,
Nika.

[QUOTE]Originally posted by loudist
From what I read there it seems that square wave response with 192k and DSD are almost identical.

This is indeed true, but the reasons we use square-wave-response as a measuring device in the analog world are clearly less relevant in the digital world. In the analog world it gives information about a combination of factors - slew rate, filtering, ripple, phase response, etc. In the digital world these indicators are much less pertinent. These are secondary characteristics that are supposed to indicate something about the performance in other ways. Digital makes the secondary performance irrelevant because the primary characteristics are unchanged even though the secondary characteristics are different.

I would take a lot of time to listen, EQ, work with, the different formats of exactly the same material.
That is the only way for it to tell the tale for me... by living with it for a while.

I definitely respect this, and I will be very curious to see what your report is. I would just hazard a preliminary guess that you determine that working with the system is somewhat inconvenient because of a lack of available processing options. I know very little about what is actually available, but I've been told it is limited.

Sonicly I have no idea what you will assess, but noticing your comments about 96k above I can't be surprised if your reaction will be similar. Whatever causes that sound for you at 96k should be present at DSD as well.

Nika.

Huh?

Oh right... nevermind... on to ignore you go!

Hello, everyone...

I'm a long time lurker, first time poster. I thought I would chime in on this thread since I have been using Pyramix at my project studio for a couple of years now.

To follow up on what others have said about Pyramix, it really is a fine sounding DAW. I might add that I don't even have the DSD option, I use it strictly in PCM mode. It is true that the non-native version is somewhat similar in architecture to Pro Tools in that it uses dedicated DSP cards for processing and the host PC is used to run the application.

I selected Pyramix as my DAW for a few reasons:

1. I wanted a PC based DAW. Since I'm a Windows network administrator by day, all of my experience is with PCs. (By the way, Pyramix runs just fine on Windows 2000; you don't have to use XP). I built a PC specifically for Pyramix and the overall system has been rock solid for 2+ years.

2. While the Mykerinos cards aren't cheap, they are plenty powerful, and I think they do more for less money than the Pro Tools HD cards (my opinion, of course).

3. Although I am primarily a tracking and mixing engineer, I do perform a bit of project mastering as well (no, I definitely do not consider myself an accomplished mastering engineer). The Pyramix software can be bought in increments depending on the feature set you need. This is really the only DAW I could find at the time which was equally at home with recording, mixing, and mastering (Sequoia may be an option as well now).

4. Though there are fewer plugins that run on the Mykerinos cards than Pro Tools TDM, the plugins included with Pyramix sound fantastic. Check out the Buss Tools for a kick ass program limiter with an excellent Auto-Release feature. The EQs are very smooth (I think Daniel Weiss may have done some of the work on the processing, though I could be wrong about that). Also, Vincent Burel makes quite a few good sounding plugins, available for purchase separately. Most DirectX and some VST plug-ins work well with the system.

5. Compatibilty is not really an issue (Pyramix can read just about any file format, including SD2 if you use a separate utility to mount a MAC drive). I suppose the exception would be full session compatibility with Pro Tools, though I believe you can import OMF sessions which are compatible with Pro Tools 4.X and 5.X (not positive about that).

Finally, I do not work for Merging Technologies, nor am I being compensated for this post. I'm just a happy user.

Thanks for listening,
Chris

Bill Dooly is using Pyramix for mastering at Paramount, and he seems to like it for that.

I have been working with Pyramix Virtual Studio since year 2000. First it was called "FrameWorks DX" and was owned by Waveframe in US. FrameWorks DX was designed by Merging Technologies - the actual owners. Since it is back in Switzerland, Pyramix is getting very close to being the musicians best solution but they do not want to add MIDI sequencing and loop playback options. Those two added in the timeline would probably make it the best DAW ever cause it's so great to work with. I work with it everyday and I'm still syncing Nuendo (only in MIDI) with it. Still it will be great to have them team up and use the Mykerinos MB6 card. Patzerat.