Rogacz

All of us have seen a lot brawls about how good 24/96 is, so after reading last part of such "discusion" I realized that it would be a nice idea, to make a dedicated topic with true, documented info about differences in bit depth and sampling frequencies.

My knowledge in this topic isn't advanced enought to post a true review with pros and cons, but I will post two interresting links concerning this problem (They both were posted originaly by igor_levicki)

1)
------
http://www.digit-life.com/articles2/...ve-audigy2-zs/

Creative Audigy 2 ZS review by http://www.digit-life.com.

Contains comparison to older Audigy 2 model, close examination, SNR marketing parameter explained and sub-article about 24 bits with it's pros and cons.

A2 ZS review may be not so much on topic, but SNR and 24 bits are, and need an example to show how things work (SNR marketing, DAC)


2)
------
http://www.tomshardware.com/consumer...editor-09.html

Tom's Hardware review of Creative Audigy 2 ZS Video Editor

Contains graphs of 24/96 performance using this card, less comments, but usable for people for who numbers do the talking.

------

Feel free to add any links or write documented info about 24/96 and comparison 24/96 <--> 16/44.1(48).
This topic will be moderated and all unnececary posts will be deleted. It's purpose is only for information, not warzone.

__________________

ursamtl

Interesting reading Rogacz. I've been fan of Tom's Hardware for years now. It's an excellent resource. As for digit-life.com, there are a couple of unknowns they fail to mention plus some curious handling of DVD-Audio. Did they have advanced speaker calibration, CMSS, etc., EAX, etc. on or off (24-bit playback won't occur if any of these are on). This is no different from kx's implementation of 24-bit audio, since the p16v handles 24-bit audio and is separate from the DSP chip, which is 16 bit. Perhaps they did have one or more of these features enabled during these tests.

The source of the "brawls" about 24-bit support in kx is not just 24-bit HQ playback since kx does in fact support this (in stereo only). what kx does not support but Creative does is 24-bit, multichannel ASIO playback. Another related feature that Eugene has already ruled out a long time ago is DVD-Audio playback which is also offered by the Creative drivers for its own DVD-Audio Player app as well as WinDVD.

This brings up another issue regarding digit-life's test. They state that they played back the DVD-Audio demo from a HDD image of the disc to avoid mechanical drive noise. Since Creative and WinDVD have some sort of agreement and licence regarding the DVD-Audio playback, it's impossible to play back DVD-Audio material at 24/96 through another set of sound card drivers. Yet, this test report claims they did so through a Revolution 7.1 card! Plus, since this report was obviously written no later than 2004, a full year before anyone cracked DVD-Audio copy protection, the disk image on the hard drive is not true DVD-Audio. Instead, it's probably an image of the DVD-Video track that is found on most DVD-Audio releases. It's basically video footage consisting of still images with audio.

What we need is a link to test results conducted on an Audigy 2 ZS with all driver updates from 2004 installed, with no DSP "enhancements" turned on and from the ASIO outs.

Eugene Gavrilov

dvd-audio playback cannot be supported, since DVD players explicitly require special 'hidden' interface in order to playback encoded 24/96 content (due to the agreement you mentioned)

kx suppots 8-channel 24/96 playback in WDM mode, btw [wave hq]

as for digit-life test, yes, there are a lot of issues when trying to test true DVD-audio quality

igor_levicki

I would add that there are CD/DVD emulators that are able to fake the drive and the CD/DVD content to any program, let alone to some rudimental DVD-A player. Regarding possibility that some option was enabled that degrades results I would not count on it too much because results are marginally better in 24/96 mode than in 16/48. If they were impaired by the DSP they would have been worse, don't you think so?
Moreover, you can clearly see the specification for CS4382 chip used in Audigy 2 ZS -- dynamic range of 114 dB (A-Weighted), and THD+N of -100 dB. No software can change those parameters which represent physical properties of the converter. The maximum recording dynamic range is approximately equal to 20 x log (2^N), where N is the number of bits. For 24 bits that would be 144.49 dB. Having 114 dB (theoretical maximum) means that you actualy have only 19 bits of resolution for playback.
That can be considered either as 18.75% of an improvement over full 16-bit dynamic range or as 26.74% degradation of true 24-bit signal -- chose the one that best fits your thinking model.

__________________
Regards,
Igor Levicki
--
Creative Audigy 2 ZS Retail (SB0350) kX 3538h

Toadly_Ossum

So what do all these number mean and how important are they?

www.tweakheadz.com has a great article here> http://www.tweakheadz.com/16_vs_24_bit_audio.htm

Demistified it for me!

Toad

igor_levicki

Great article, confirms my thoughts completely. In short, without technique and talent 24/96 is useless.

__________________
Regards,
Igor Levicki
--
Creative Audigy 2 ZS Retail (SB0350) kX 3538h

ursamtl

There are two sides to every story. To quote well-known recording engineer Bob Katz:

The first requirement for good sound is 24-bit data storage and processing. If your workstation only deals with 16-bit words, then all you should do is edit. Don't try anything else, unless you're actually looking for that "raw and grungy" quality. Do you want your music to lose stereo separation, depth and dimension, become colder, harder, edgier, dryer, and fuzzier? If these are not problems for you, then keep on bouncing and mixing away in 16 or 20 bits.

from More Bits Please! by Bob Katz.

I'll let you read the rest of the article. By the way, I've stated before on the forum several times that it's not the 96kHz part of 24/96 that interests me. One of the most frustrating parts of waiting for kx to implement the 24-bit ASIO outs is that Eugene has the control panel switch in there for 24/48, which is to me the best solution. Most audio experts I've read have stated that bit depth is far more important than sampling rate once one gets past 44.1/48k. Creative only offers 24/96, which I feel is a needless waste of HD space.

ursamtl

Yes I was aware of that from one of your earlier postings on the subject.

When I ran kx on my A2ZS, I was under the impression wave hq was only 2-channel stereo. Where would the output connections be in the DSP window?

ursamtl

As Eugene just mentioned, DVD-A cannot be supported because of hidden interfaces. This format is heavily copy protected with several layers of protection. (which is why the industry freaked when cracks appeared last month). Therefore, successfully doing an ISO image of a DVD-A is unlikely. The data might be in the file, but it would be unplayable. The audio track in the Video_TS folder would be playable and would fool someone into thinking he or she was actually listening to DVD-Audio. For some details, check out http://www.disctronics.co.uk/technol...d_copyprot.htm

igor_levicki

Quote:

Originally Posted by ursamtl There are two sides to every story. To quote well-known recording engineer Bob Katz:

I will just comment on the part of the quote you selected from that article:

Quote:

Originally Posted by Bob Katz Do you want your music to lose stereo separation, depth and dimension, become colder, harder, edgier, dryer, and fuzzier?

How in the world you can lose stereo separation by using fewer bits?!?

How can something sound colder, harder, edgier, dryer, and fuzzier at the same time?!? Cool and dry -- that I can understand but if something is hard and it has edges how can it be fuzzy at the same time?!?

Moreover, I agree that each bounce (like in "bounce to track") and every other unnecessary edit operation has detrimental effect on audio quality but it is the same for 24/48 or 24/96 -- those formats are not lossless compared to 16/48.

Also, I will quote something for you from that article that proves one of my points:



Unfortunately, seems that I have lost my DVD-A demo disc that I got with Audigy 2 ZS. I cannot remember if it was protected or not. From the screenshots it is obvious that they were playing that disc. Since I don't have it anymore I can't try to do the same thing they did.

__________________
Regards,
Igor Levicki
--
Creative Audigy 2 ZS Retail (SB0350) kX 3538h

ursamtl

Igor, you're taking these adjectives literally. In English when these are used to describe sound, what Katz says makes perfect sense. Besides "edgier" does not mean that it has distinct "edges." Rather it means that something sounds more harsh or irritating.

Perhaps you could write Bob Katz and ask him. I tend to take the guy seriously given his experience as a recording pro. Check out the list of people he's recorded over the years:

Anytime there is a conversion or calculation there are going to be losses. It's just that there are many more losses at 16 bits than at 24. Igor, I think you need to do some research concerning audio mixing and production. It's well known that performing mixing, etc. with 24-bit material is far less detrimental than for 16-bit material. You should check out the Dithering with Ozone guide. This provides one of the clearest explanations of dithering I've ever seen. It explains quite well the difference between 16- and 24-bit audio.

As for your quote from Bob about it the need for "a very special digital console to preserve 24-bit quality" well yes, I agree that a commercial sound card from Creative Labs is not going to do that as well as high-end equipment. However, it's still going to do it better than the same sound card at 16 bits. Even if the improvement is only the equivalent of 19 bits, it's still an improvement applicable to the hardware one already has in his or her computer. kx has never been about making one's sound card do things beyond its capability; instead, it was about maximizing the potential of the things the card was capable of doing but Creative's driver programmers ignored until last year.

Look Igor, you can go dig out articles to back up your case. In turn, I can go dig out articles to back up my case. I don't think either of us is going to change the other's mind. Let's just agree to disagree.

TravelRec.

Sorry, guys - all what you say is true for sure, but for myself my ears are detecting the sound and if they find it sounds cool, there´s no need for me to check whether this sound comes out of a 16/48 or a 24/96 machine. More important is the noise you will record with the signal, and that is recorded on both machines if you have a bad analog input signal from your mic or your synth or whatever. If you try to get the signal as clean as possible before you feed it into the card, then you are one step ahead. The mix setup (in the sequencer) is another important thing. I think (my opinion) that the equipment outside of the computer (amps, cables, mics, sockets and plugs...) is pretty important, to get a fine signal. Some of this stuff is relatively cheap to upgrade. If you want to have a good 24/96 digital equipment, you have to pay A LOT for maybe subtle results...

__________________
Always take a look behind.

ursamtl

Everything you say here is also true, if we already have a piece of gear in our system that support 24-bit sound, a subtle improvement at no cost at all is worth it! Again, it comes down to getting the best performance out of what we already have.

nakamichi

I'm posting this a bit late,but it's still a good read for all of you who think Audigy2 is really 24bit:


24/96 vs 16/48 difference with Audigy2 ?
----------------------------------------
Benefits of 24/96 mode:

- Extended hi-frequency response and flatter frequency response curve
- 6 dB better Dynamic Range,Stereo Separation and SNR
- Half the level of THD as in 16/48 mode
- More bits of the actual sound reproduced (18 bits instead of 15.5bits) = a bit smoother sound

SB Audigy 2 is an 18-bit card,not 24-bit,live with this fact. But 18-bit sure sounds better than 16bit,although the difference is very subtle.
Audigy2 and audigy2 ZS have the same specs, ZS is NOT better in terms of sound quality than plain Audigy2 Retail,just offers one more extra channel.
Audigy2 Bulk is of slightly lower quality though.
Even the best E-MU cards are still 20 bit at most.No true 24-bit card is available for now,even the best pro cards are far from 24bit.

I myself can hear a difference between 24/96 and 16/48,noticeable when creating a project in a sequencer and using high-resolution effects mixed in with software 32 or 64bit float precision math engine.
I switch the ASIO from 16bit to 24bit and vice versa and compare the sound quality (with ASIO4ALL)
The difference is not dramatic,though-it's very subtle.It sounds like just 10% better,not much but still noticeable.Sounds like going from 16bit to something like 18bit,not to true 24bit.And this is exactly how much better is an Audigy2 card from a high-end 16bit card.
The sound is a bit beefier,effects are a bit more accurate and it has a bit more "air".Not much.I don't think even the newest upper-end X-Fi series will come close to 24bits.Maybe some of the later,most expensive E-MU X-Fi based cards will reach that level...I can't wait to buy a TRUE 24-bit card at a reasonable price.
For a true 24-bit experience you need an extremely high-quality receiver or a soundcard with real 24-bit DACs,and ADCs if you record too.
16-to-24bit difference in this case is really dramatic.With Audigy2 it's just subtle.It sounds like souped-up 16bit,not like the "real 24bit thing" yet.
And yes,analog is infinite resolution,but with other flaws,like low frequency response,a good deal of noise,high THD and playback speed variation.
...but analog sounds closer to the real thing,no matter how big these other flaws are.
I still listen to cassette tape after all these years of DVD-A...I won't leave the Nakamichi cassette deck anytime soon


Oh,wait - what about this situation : you stream from the Audigy2 SPDIF OUT to a high-quality receiver with really good DACs,bypassing Audigy2's much lesser-quality convertors and THEN compare the difference in quality?
So,when playing DVD-A the card shuts down the SPDIF,so the solution is to play YOUR OWN MUSIC from a sequencer and feed the spdif digital data to the receiver.Then the SPDIF is not blocked,I think.
In this case 24/96 is very useful as well.

I also am one of those who think quality doesn't matter that much if the music you listen to is well composed,with a brilliant arrangement,carefully balanced,has many hidden layers of depth,harmony,the melody is catchy,the bassline and rhythm are spot-on,etc.
The "soul" is what makes the music.Quality is second priority.I'd rather listen to a great piece as a 32kbps 22kHz crap-quality MP3, than something in full 24/192 DVD-A quality that has no sense of groove and dynamics,is repetitive and sounds like a computer has generated it automatically without a man telling it what do do
Like when I do a real translation vs. a bad Japanese-to-English translation from Babelfish.
That's why I still listen to crap-quality normal position cassette tapes erased 100s of times over and over,it's the stuff that's there is important,not the quality.
But,if you give me to choose of a lower and higher quality version of the same thing - quality matters
I should've waited to buy a decent E-MU soundcard,not jumping the gun and buying an Audigy2 when it was first released,where there were no E-MU cards of the same class
But,if I didn't buy an Audigy2,I wouldn't knew of these wonderful drivers that beat anything pro out there in terms of flexibility,capability,speed and efficiency.

I wanted to get a good 10k2 E-MU card,but now I'm waiting for an X-Fi based E-MU card instead...but it's very hard to leave the kX drivers,once you have felt you had the control.

??? Isn't everything being processed in 64-bit floats now,not 32-bit? Cubase SX,all those Voxengo products,even Trackion2 does that.FLStudio seems to be the only thing left with "only" 32-bit float processing.
I hope they will overhaul FLStudio soon as well.Everybody uses FLStudio as a VSTi in their big sequencers,pros like it even more,but refuse to admit - it's a well-known secret.

I think support for more kX-based soundcards is more important for now than 24/96 ASIO stuff. More cards supported = more people using and enjoying kX drivers. SB0240 is broken in 3538i,needs to be fixed ASAP.
If you had an Audigy2 Value and your card was not supported,how you would feel?

Peace

ursamtl

Nakamichi,

Another interesting, informative post. I wonder, however, about the statement that the Audigy 2 is an 18-bit card. I know you probably arrived at this conclusion by comparing its SNR to the theoretical maximum od 144dB, but if you Google 24-bit ADC, DAC, etc., you'll find that lots of devices are called "24 bit" and their SNR normally falls within the range of 100-110db. So in comparison, our poor old Audigy 2 doesn't do too badly! Here's an interesting link to a listing at www.musicalfidelity.com for what they consider to be "just about the best DAC available in the world at any price." Its SNR is 120db, well below 144dB. I know this could simply be marketing hype, but given that most of the devices I mentioned above are below this, obviously, other real-world factors affect the SNR besides a 24-bit data wordlength. The fact remains that the data entering the DAC is in a 24-bit format, which means the calculations will be working with more sound information than if the data entering had a wordlenght of only 16 bits.

Much of this debate also comes down to personal preference, with mine being fidelity. Over the past few years I've found myself becoming more and more concerned with fidelity. Even subtle differences are important to me. For instance, on my work computer, I have a half-decent set of Koss headphones hooked up to its onboard Soundmax card. I was astonished to discover how much distortion disappeared when I started running Foobar2K with SSRC resampling everything to 48kHz. The soundchip obviously isn't well designed. The improvement was somewhat like the quote I posted yesterday from Bob Katz. It was less edgy, more natural sounding. Although at home my A2 doesn't show as much difference, there is a noticeable improvement as well. Like you, I also notice a similar subtle improvement when I listen to 24-bit audio through my system at home. It's not quite as dramatic as the 44.1/48kHz difference on my work PC, but it's still there.

As far as the "soul" of the music goes, I understand and appreciate what you're getting at, but I find my listening experience is greatly enriched by fidelity. For example, I'm a huge fan of classic Motown, especially old Four Tops and Temptations hits. When remasters of these became available, I rushed out to get them. Unfortunately, I find myself disappointed by the quality. Levi Stubbs was an incredible vocalist whose performances to me exuded more "soul" than almost any others, but the distortion on his voice ruins it for me. The mono releases are marginally better but I hate the one-dimensionality of mono. I'm still hoping Motown will release clean, stereo remasters someday.

Finally, I do understand the need for support for certain trouble cards. Eugene has made it clear now that the coding effort required to implement 24/96 ASIO is much more than for fixing these issues. Up until his statement this week, I was under the impression (at least partially created by those who took it upon themselves to speak for Eugene) that the 25/96 ASIO stuff was being sacrificed just because a few people complained that their card wasn't supported. I now see this is not the case. I also explored ASIO4ALL and discovered that the latest versions do in fact support multichannel playback. I haven't had time to remove all the Creative stuff and put kx back on yet, and to be honest, I'm still undecided as to whether I will. In the past few days I've thought about this and realized that I've grown fond of the DVD-Audio playback capabilities, and the simplicity of the "record what u hear" feature (I know it can be done with kx, just not as easily). I'm also waiting for a reply about the modded A4 driver, to see if it will give my A2ZS anything beyond what I have now.

Regards,
Steve.

igor_levicki

Steve, you are mixing SNR and dynamic range. SNR (or signal to noise ratio) for 24-bit ADC/DAC is ~110 dB as you just said, but 144.5 dB figure you are comparing it with is teoretical dynamic range for signal with 24-bit resolution. Maximum dynamic range that human ear can perceive is ~120 dB.

About mixing, I understand it very well and I know the math. As long as the signals you are mixing do not exceed the 16-bit dynamic range you are fine. After that you have clipping and distortion. By doing careful mixdown you can always have clean and great sound even with 16 bits.

About internal precision of programs doing audio processing, some programs indeed work with 64-bit precision. Problem is that many effects plugins perform processing in 32-bit precision to be able to work in realtime. Remember that the final sound always has the quality limited by the weakest processing chain.

Also, it would be better if the sound was processed in fixed point because floating point format introduces artifacts too, read about it here.

About X-Fi, 24/96, dynamic range and Creative "cutting corners" strategy you can read here.

__________________
Regards,
Igor Levicki
--
Creative Audigy 2 ZS Retail (SB0350) kX 3538h

ursamtl

This is more a debate of semantics than anything else. Dynamic range is normally defined as the difference between the softest and loudest perceived sound levels a device can produce. Signal-to-noise ratio is the difference between the noise floor of a device and its maximum sound level. Since anything below the sound floor is going to be unheard, the "practical" dynamic range of a device is going to be the same as its SNR.

Besides, I'm not the one using 144.dB as a point of comparison. You and others have complained about the A2 not being a "true 24-bit" card because it doesn't produce anything near 144.5dB. For example, yesterday you wrote:

Moreover, you can clearly see the specification for CS4382 chip used in Audigy 2 ZS -- dynamic range of 114 dB (A-Weighted), and THD+N of -100 dB. No software can change those parameters which represent physical properties of the converter. The maximum recording dynamic range is approximately equal to 20 x log (2^N), where N is the number of bits. For 24 bits that would be 144.49 dB. Having 114 dB (theoretical maximum) means that you actualy have only 19 bits of resolution for playback.

Yet today you say that the maximum dynamic range that human ear can perceive is ~120 dB, so what's the point of having the extra 20dB of range that no one can hear?,Also today you say a "24-bit ADC/DAC is SNR ~110 dB". Since the A2 has a rated SNR of ~108dB, I'd say it qualifies as a 24-bit device! Yes you might produce links to reviews that rate it at less than 108dB, but then I can produce one that doesn't. Sound on Sound magazine is one of the most respected music recording magazines in the world. Here is what they had to say about the A2ZS.

Rogacz

Please include pro reference if writing controversial things.

I just came with the idea:
Why not make a pool concerning 16/44.1(48) or 24/96. Do you hear the difference or not. Maybe not so good question, but everyone should get the point. This would show us some data about this topic in real life.

__________________

ursamtl

The problem is that what may be controversial to one person is common knowledge to another. For instance, if you're referring to my definitions of SNR and dynamic range, these are concepts I learned years ago.

Listening polls can be very tricky, misleading. You'd need to define the conditions, and ensure it was done at least with some attempt at a scientific approac, probably double-blind testing. This is why hydrogenaudio has such stringent rules concerning quality statements.

nakamichi

ZS has rating of 108dB SNR? C'mon - that's just a marketing trick. It was rated at 106dB,and at the last minute,they changed the label on the box,claiming that now it is 108dB.Bullcrap,it's the same chip and DAC.
I think,it's not even close to 106 in real situations.
Dynamic Range and SNR are different things,you may theoretically assume in this case that it's the same,but Dynamic Range is practically lower than the SNR the card is rated.
In this case of the Audigy2(ZS),you might get better SNR and Dynamic Range if it had better DACs,but as the situation is now,it can hardly reach 100dB,and that is when using the SPDIF out.
Dynamic Range and SNR of over 120dB IS useful,for purpose of mixing to minimize noise and get the resulting noise floor to under 120dB,so it will be perceived as totally noiseless to the listener.
In fact,the human ear is capable of going up to the maximum theoretical dynamic range 24-bit audio offers,just anything over 120 is damaging to the ear,so it can be used ONLY for very short
instances of LOUD noise (explosions,gunshots,airplane engines and such) to simulate how it is in real life. These types of sounds are really loud and easily reach the upper floor of the 24bit range.
There should be a warning,like: WARNING - 100% realistic simulation of Explosions and gunshots can lead to a temporary or permanent hearing damage,so listen at your own risk,if your sound system can support it and you are willing to turn up the volume so the full dynamics can be reproduced.
Instructions: set the volume to a very loud level,stay 20cm away from the speakers and turn up the volume until you can listen to a very quiet whispering sound.Then move at least 5 meters away and cover your ears,and see if it will produce a very loud sound.If the sound is distorted,your system is not capable of sustaining the immense sound power.
If everything is correct,your volume is set at the right level,so you can "enjoy" and take the risk.We take no responsibility of your actions.If you damage your ears,it's all YOUR fault.

ursamtl

Marketing trick? You obviously didn't read the link I provided above.

The Audigy 2 ZS Platinum Pro certainly passed the RMAA 24-bit/96kHz tests with flying colours, confirming the 2dB improvement in dynamic range, a slight improvement in stereo crosstalk, and a more extended high-frequency response, being 0.1dB down at 45kHz rather than the 1.0dB I measured on the Audigy 2, showing that a different filter characteristic had been used. As on the original, however, there were still characteristic ripples at the high end at all other sample rates, where ASRC (Asynchronous Sample Rate Conversion) had been employed, although once again the dynamic range was slightly better.

From review in Sound on Sound, October 2003 issue.

If you're basing your comments on the digit-life report, as I wrote yesterday, it seems rather suspect. They don't tell us about their test conditions at all. If you want more transparent test results, why not check out http://audio.rightmark.org/? You can get their own Audigy2ZS Platinum Pro measurements. You can also get Creative's own Audigy 2 ZS Series Testing Methodology from Creative. It describes the test methods used, etc., in a surprisingly non-marketing hype way.

nakamichi

Marketing trick: they improved the SNR by updating the drivers,not by changing/replacing any hardware.It's just the drivers.
And Creative is well known making drivers that are not equal for all cards,artificially limiting certain cards just for making more $$$ by forcing users to buy the more expensive versions.

I said: compare them both (A2 and A2ZS) with the same drivers (kX) and you'll get pretty much the same results

ursamtl

I agree that the big deal about upping from 106 to 108dB is marketing "hype", The term "trick" is not the best choice since it implies blatent dishonesty. Obviously under some circumstances, Creative was able to get results of 108dB, so they decided to push that aspect. Although Sound on Sound magazine didn't publish their complete test results, they did state in their review that they confirmed the 2dB improvment in their tests. Rightmark's own test, for which I provided a link last night, resulted in a -107dB noise level under DirectSound. A difference of 1dB is certainly within a reasonable margin of error. Both of these independent results confirm that the A2 is not some substandard product whose manufacturer is trying to hide real dynamic range and SNR figures that are well below the competition. Yes, they might be well below some theoretical maximum that's either unattainable or else only possible with prohibitively expensive equipment, but in terms of similar products, they're ok and certainly better than the A2ZS's own 16-bit results!

For the purists, you might consider automobile gas mileage ratings. The official figures are always attained under the most ideal conditions possible. It's virtually impossible to obtain such results in real-world situations. They are simply for comparison between models in a given class or price range. Thus, if you compare a consumer-class soundcard with pro equipment costing thousands of dollars, the poor little card is going to look like absolute crap!

Besides, the CL notice displayed on thd digit-life page makes no mention of any hardware change resulting in this improvement, nor does it imply any reason. It simply states that, at the last minute, they were able to improve the SNR by 2dB. I personally think this magnitude of SNR improvement is mostly irrelevant since the difference would be basically undetectable to the average human ear. Perhaps using a different driver would make a difference of 2dB, but if it didn't provide functionality I wanted, I'd put up with a 2dB drop in SNR for the increased functionality.

As a technical writer, I've had to work with marketing people frequently, and this kind of selective emphasis occurs a great deal of the time. I always try to remind such people that not all customers are uninformed and some will see through such hype. However, let's face it, the primary goal of commercial enterprises is wealth generation (i.e. selling their product so they can stay in business), so if taking a fact and over-promoting it will help, then it's going to happen. CL is competing with other soundcard manufacturers, so these specs are looked at by a lot of people when deciding which card to buy. Granted, only some complete techno geek without clear knowledge of audio is going to base his choice solely on a 1-2dB difference, but if it makes another sale for the company selling the product, emphasizing the spec can't hurt. I think it's more important to note that in the comparisons reviews I've read, the A2ZS stands up very well in comparison to its competition when it comes to performance.

Don't forget that sound quality is only one aspect. When I purchased my A2ZS last December I might well have chosen other cards such as the M-Audio Rev. 5.1 or 7.1 for example. However, for my usage, I need MIDI input/output, preferably 2in/2 out. I would have had to buy a separate interface for this had I gone for the M-Audio product (notice I didn't say "solution." That's a marketing buzz word that's been so grossly overused it makes me want to vomit every time I hear it). The A2ZS gave me the MIDI ports I needed, specs very similar to the M-Audio cards, DVD-Audio playback, a decent software package, etc. As I mentioned in other threads, I also based my decision on Eugene's mentioning that he was working on 24/96 ASIO for kx. Unfortunately, that didn't happen so I ended up going back to Creative's drivers. Nonetheless, I'm still quite happy with the performance of my card. Obviously others on this forum are too or we wouldn't own the card or be here discussing it!

igor_levicki

It is true that noise will mask useful signal, but there is also digital input and output which should be unaffected by this noise, right? So after all, SNR and DR can't be mixed like that. If you still insist on those two terms being interchangeable then let us just say that you have even worse numbers to work with -- theoretical 108dBA SNR (claimed for analog output at 2V signal with AES17 filter -- read: not in common use). In practice (measured with RMAA at digit-life) you have -97.3 dB noise level for 24/96 and -101.1 dB (slightly better) for 24/96 over SPDIF. That still falls short from marketed 108 dBA SNR.

Moreover, system noise is usually measured without signal (i.e. input shorted to ground). That means you will get zero samples when recording from that input and after 8192 zero samples played, CS4832 used in Audigy 2 ZS turns it's DAC completely off (auto-mute) thus producing results close to theoretical maximum SNR. I am inclined to believe in those measurements made by folks at digit-life since they were aware of that problem and they have found a way around it or otherwise the measurements would be pretty useless. After all, they wrote the RMAA which requires considerable knowledge on the subject.

I was quoting you when I said that:
I suggest reading this article which talks about digital audio and common misinterpretations.

Also, I recommend reading all articles in ESP Column, you would surprise how much fraud is floating in hi-fi waters.

__________________
Regards,
Igor Levicki
--
Creative Audigy 2 ZS Retail (SB0350) kX 3538h

ursamtl

Well why don't you check what the folks at digit-life themselves said:

Besides, we should have done this under the same conditions, as Creative's engineers had.

They admit that their tests were not done in the same way. And what about the rightmark.org tests I provided yesterday? You ignore them yet they also wrote their RMAA test "which requires considerable knowledge of the subject." Why are their own test results so much better than digit-life's? Perhaps they followed Creative's instructions for testing.

Seriously, why don't you go read my post this morning. The point I make about these numbers being meaningful only as a comparison between similar products is, I feel, I strong one. There's no point in comparing them against unattainable theoretical maximums because they'll always fall short. There are lots of tests and reviews on the internet that place the A2ZS at or near the top when compared to other similar products.

As for your links. the first one is just another of many sites with information related to recording. Most of it is presented fairly well although some of the writing is a bit suspect. They attempt to say that recent tests of DACs found that many only "captured 17 to 19-bits of sound." However, sound is analog and capturing it involves running an analog signal into an ADC, not a DAC!!! then later on the same page, they make the following statement:

If you wish for a high-fidelity alternative to 16 bit 44.1 khz, it is better to increase the bit depth (to 24) than to increase the sampling rate.

I couldn't agree more. I've read about the importance of bit depth on virtually every pro site and in every pro document. It is far more important than sampling rate. But of course, if the devices can't really do 24-bit, as they imply, how can moving to 24-bit be a "high-fidelity alternative"?? You can have it both ways!

As well they suggest that the increased clarity attributed to recording at 96kHz can be "simulated with a positive high shelving filter." In other words, studios all over the world can throw out those expensive 24-bit devices and just boost the highs for more clarity!

I could go on, but you get the picture. It's not a bad site, just a bit suspect in spots.

I also know the ESP site as I've checked out some DIY circuits there before. Look, it's a company web site with products for sale. Of course they are going to rave on about the bad industry to make their products look good. I'm sure some of the stuff is true. Creative is going to promote their products. Likewise M-Audio. Ditto Terratec.

By the way, according to your signature, I see you have an A2 ZS. If you think the card is so lousy, why do you have one?

igor_levicki

Of course they do because it is not possible to repeat tests if you don't have exactly the same equipment and you set it up exactly the same way.

IMO they are the only reviewers who had detailed Creative measurement specifications. I haven't seen anyone else bother to even come close to reading those specifications before writing a review let alone trying to interpret and meet the numbers provided by Creative.

I do not ignore them, I told you that guys from digit-life are the autors of the rightmark audio analyzer so you can't chose to trust them at rightmark.org and not to trust them at digit-life.com.
I don't find it drasticaly better. In MME mode, they have measured 99.3 dB while digit-life got 97.3 dB. And the main thing you are ignoring is that digit-life tested A2ZS retail and rightmark test is for Platinum Pro which has better SNR because it has ADC outside the computer feeding the card digitally. So please read more carefully next time.



It is obvious that they have made an error. But their points are still valid:

1. ADC (and DAC chips too!!!) are advertised and sold as true 24-bit when in reality they fall short from that resolution. Way too short.

2. It is better to have HQ 16 bit ADC/DAC than LOWQ 24 bit ADC/DAC.



Sigh, it can be "high-fidelity alternative" when you have HQ ADC/DAC. Now we just have to agree on term HQ here and to end this useless debate. For me ADC and DAC circuitry in Audigy 2 ZS series are not HQ but I can't afford better at the moment and that is also the answer to your question about me having A2ZS.

Quote:

Originally Posted by ursamtl I could go on, but you get the picture. It's not a bad site, just a bit suspect in spots.

Suspect in the same way Bob Katz can lose stereo separation by using fewer bits? Come on...

Quote:

Originally Posted by ursamtl Look, it's a company web site with products for sale. Of course they are going to rave on about the bad industry to make their products look good.

Again you are not reading carefully. I wasn't referring to their products but rather to the tech-savvy analysis of some of the most common frauds in hi-fi.

Listen Steve, no hard feelings but to me you seem to be the type of person who would buy 3,000 USD golden mains cable for power amplifier because it "sounds better" that way. I wouldn't because I have been trained over the years not to be a consumer and not to fall for marketing hype. That means we will never agree. You believe that the minor improvement is worth the effort (or money) even though you can't prove that it is indeed an improvement and I don't. Period.

My father was into DIY when neither you or me was born and when electronics was all tubes. Just like Elliot, he has a great deal of knowledge and advice for those who would listen. I would trust him or Elliot anytime because they are not selling me anything and neither is digit-life.

Btw, check this and see for yourself how close are those new 24/96 integrated(!) AC97 chips (Azalia HD Audio) to the A2ZS measurements in RMAA. Shurely shome mishtake? I wouldn't say so. And they cost nothing compared to A2ZS. Also, check ESI Juli review and comparison with A2ZS Platinum Pro and Lynx Two, you will be surprised. Now that is the card I would like to buy instead of this junk I have now.

__________________
Regards,
Igor Levicki
--
Creative Audigy 2 ZS Retail (SB0350) kX 3538h

ursamtl

Igor,

I told you a couple of days ago that this was a useless debate. I'm glad to see you're finally seeing that too. It's just too bad you feel some need to continue it! Do you see this as some sort of competition? Where's the prize? The only reason I don't ignore you completely is because if others who read this believed everything you wrote without it being challenged, they'd run away with the idea that their money was wasted on some horrible "piece of junk" (as you call it)!

Of course there are pro and even semi-pro products that can best the A2ZS. We all know this. However, with the right driver (kx for 16/44; CL for ASIO 24/96) and a proper set up, it is far from a piece of junk. Numerous reviews and tests have placed it at or near the top of its class.

As for some of your other statements, let's just say you obviously don't understand how stereophonic sound is produced and propagated if can't understand Bob Katz's point about bit depth and sampling rate affecting stereo separation. Are you saying that a widely known and respected audio professional is wrong and you're right?

If you check Elliot's site, you'll find he is trying to sell a product. So is digit-life. It's called advertising space!

Yes, the ESI Juli is an interesting card, but it only exceeds the A2ZS specs by any significant degree when one uses the SPDIF outputs. In such a case, one also has to purchase an additional external DAC, and one of very high quality in order to benefit from the Juli's better SPDIF specs. Otherwise, when using the card's own DAC, we might as well stick with our A2ZS. Plus, although its software supports 5.1 output via SPDIF, its analog outputs do not. The lack of 5.1 analog output is not clearly stated in this review, but 4-in, 4-out would seem to suggest that. In fact, the digit-life review does warn that home users who use the card for high-quality playback would be better off considering mastering cards.

As for presumption of my cable buying habits, actually I'm the type of person who doesn't buy gold-plated cables because their minor improvement in specifications is not audible! The same way those obsessing about a dB or two difference between sound card performance results are similarily misguided IMHO. Were I running a pro studio, I'd want the best specs possible within whatever budget I had, but for home studio use, the A2ZS is a good choice, as lots of the forum members here have demonstrated with their projects.

Rogacz

I asked you guys to insert only info about 24/96, not to make a warzone.

Please calm down, and if you see, that this debate has no sense, insert some sort of post with pros and cons of using 24/96 or lower modes in points, or send it to me, so I can add them to first post.

__________________

ursamtl

Some have already provided some good points as well as pros and cons. The tweakheads article provides good advice concerning tthe benefits to sound quality of 24-bit recording. Certainly the Bob Katz article I provided provides the view of a seasoned, experienced recording professional. The dithering with Ozone document link I posted also provides tremendous information. As for making points, I've already said in this thread as well as others that from all the reading I've done on the subject, it has become clear that a bit depth of 24 is far more important than 96k sampling rate.

By the way, it's not about creating a war zone, it's about making sure erroneous comments aren't left unchallenged. I've made a couple of errors and Igor pointed them out, just as I pointed out his errors or errors I found in the digit-life article. The war zone comes up when comments along the lines of "you seem to be the type of person..." come up. This is personalizing the debate and criticizing the individual instead of the idea.

As for the original subject of 24/96, I made up my mind awhile ago (24-bit good; 96k unnecessary), so I really don't see the point in my continuing to participate in the debate.

Rogacz

That's what I'm talking about.
Let the topic be about 24/96, not personal likings/dislikings.

Just make this thread a meal for everyone who wants to know important things about 24/96, not a personal topic.

That's the point.

__________________