[kartik]

Quote:

Originally posted by Orpheus yes, there's probably empirical evidence either way. but i'm addressing those that say "SACD 'sounds' better." i am saying, you can't say something sounds better until you can identify the differences when comparing the same material. i don't understand how people can say one sounds better than another when comparing apples and oranges.

Do you mean assertions that SACD sounds better than DVD-A or the comparison with redbook? IMHO the comparison with redook is a favorable one, but I will never know about the DVD-A onebecause I have no more money left!

 

[theaudiohobby]

Quote:

to fairly compare SACD to DVD-A, you must use the same recording, same processing, same level of convertors, and be monitored through the same equipment and speakers. if you do not, it would not be a fair comparison. and before you do such critical listening

Orpheus,

Did you read the dcs paper at all, because that is exactly what DCS did, they have a schematic in the paper showing the demonstration setup. And their findings are similar to the philips' paper on the subject.

 

[Orpheus]

oh okay... then i should read it. thanks. that's exactly the info i want....................... but unfortunately i'm in china right now bumming off a friend's computer so i gotta be quick. i'll look it over when i get a chance...

but can you just briefly summerize for us what it says in a couple sentences?

 

[Joe Bloggs]

Quote:

Originally posted by theaudiohobby Joe, most of your post is simply not so, but rather than get into what is and what is not read this arcticle from optical-disc systems journal, remember that is a partially a critique of DSD and you will see that the experts give more due to DSD that you want to accept. DVDA v Bitstream article also look at this article from this dcs paper and this paper and see that there are valid scientific reasons why folk might prefer DSD to DVDA happy reading.

Next time you might want to, oh I don't know, *read* your 'references' before posting them, and see whether they even support your point of view.

The DVD-A vs Bitstream article is flatly against SACD, listing every negative I've put out and more besides:

-inefficient use of bits (and consider that its bitrate is only slightly higher than 24/96 and absolutely can't compare with 24/192!
-hard to play back properly (nonlinearities caused by the many noise-shapers in the playback chain)
-can't be used for digital processing, hence must be converted to PCM in studio and back to DSD before release (although with the master tapes now being 8-bit, maybe this is avoided?)

If you don't understand anything else in the article, let me quote its conclusion:

Quote:

We consider that these advantages of PCM far outweigh the basic advantages of bitstream and we therefore recommend a losslessly-packed linear PCM system to you for formal adoption.

The dcs paper may seem to be in favor of DSD, but on closer inspection, for every criterion where there was a negative observation for 24/192, there was NO OBSERVATION made with DSD yet. It's 'yeah, we know there's a bit of a problem there' vs 'we don't know how it's like yet'. How can this be taken as support for DSD?

Including the last link as reference is plain ridiculous, as there was not even a single mention of DSD or SACD in the whole paper.

Are you trying to waste my time?

I don't know--maybe you forgot to link to the part of 'this arcticle from optical-disc systems journal' that favors SACD. But right now it looks like your research sucks.

 

[theaudiohobby]

Quote:

Conversion to a 1-bit code (typically from a 4-bit code) is then performed by an open-loop, optimal code conversion table which minimises spectral modulation. Potentially the system produces high resolution with no low-level correlated distortion. ... We recognise the advantage of bitstream in a basic system and the natural extension of ultrasonic bandwidth ...

That is a quote from a paper that is dead set against bitstream encoding, the point of getting you to read the paper is to point out to you even the opponents of bitstream acknowledge its superiority in certain areas. The DCS papers infer that DSD in controlled listening conditions is superior to 192/24 and secondly that time resolution is proportional to sampling frequency. As I consistently said in this thread, I am yet to read anywhere that anybody has disputed the superiority of DSD time resolution over 192/24. Now that you have read that I think it is time to read response . The conclusions are clear that infact SACD is more bit efficient than DVDA and has better time resolution and has a more linear response for a given input. The second paper discusses issues of transient response that is key to understanding DSD's superiority over DVDA and one of the key reasons why slowly but surely DSD is pulling away from DVDA. There is more to music than just bit resolution and that realisation is what is finally causing digital to pull away from vinyl based systems. One thing that a digital system should have over vinyl based systems is precision, however if your precision is focused on the wrong parameters, then you will have missed the mark by a long shot and I submit to you that SACD has made the right choice but choosing to preserve time resolution over its useful bandwidth.

Once again happy reading .

 

[Joe Bloggs]

Quote:

Originally posted by theaudiohobby That is a quote from a paper that is dead set against bitstream encoding, the point of getting you to read the paper is to point out to you even the opponents of bitstream acknowledge its superiority in certain areas.

That quote is also only relevant to the 'alternative bitstream converter' and we have no idea whether such an alternative is being employed. The 'advantages' in that quote also happen to be that this methodology solves inherent problems in DSD that were never present in PCM to begin with!

Quote:

The DCS papers infer that DSD in controlled listening conditions is superior to 192/24 and secondly that time resolution is proportional to sampling frequency.

That's a flat lie. This is all that had been said about DSD in the paper:

Quote:

44.1 kS/s, 16 bit (current format) 96 kS/s, 24 bit · much less “busy signal” break up – very good quality · better separation of reverberation/room acoustics from instrument output · bass better balanced · percussion (particularly cymbals) better · some stereo image formation 192 kS/s, 24 bit · no “busy signal” break up – excellent quality · very good separation of reverberation/room acoustics from instrument output · but - bass can appear light and slightly out of time · and - stereo image can be strong but widened (1.5 times?) DSD · detailed comparisons not yet performed on enough systems, but well liked by (classical) artists after sessions · no “busy signal” break up · very good separation of reverberation and room acoustics · no observations on bass so far · strong stereo image formation, no observations on width so far --- When played back on a system with adequate hf tolerance, the results from DSD are very good. We have not so far done enough comparative listening tests (DSD vs PCM or DSD vs Analogue) to comment usefully on the differences. We have done some testing of different filters in the noise shaping loop in a DSD modulator. However, even these results are dependent on the monitoring system and useful comments cannot be made at this stage.

They explicitly stated that no useful comparisons between DSD and the PCM formats! As I said before, where deficiencies in 192/24 were noted, no corresponding observations had even been made about DSD! 'Time resolution is proportional to sampling frequency'? Where did they say anything like that?

Quote:

As I consistently said in this thread, I am yet to read anywhere that anybody has disputed the superiority of DSD time resolution over 192/24.

And *I* have yet to read anywhere that anybody has stated DSD time resolution to be better than 192/24! So what if nobody spoke against a claim, if that claim had never been made in the first place!

Quote:

Now that you have read that I think it is time to read response . The conclusions are clear that infact SACD is more bit efficient than DVDA and has better time resolution and has a more linear response for a given input. The second paper discusses issues of transient response that is key to understanding DSD's superiority over DVDA and one of the key reasons why slowly but surely DSD is pulling away from DVDA.

1. There is only one link here, to the content page. I assume the first article I'm supposed to be looking at is 'SACD Format: Development of the DSD Recording Solution'. Where's the second? Or is this the second?

2. Stop putting words into people's mouths (or paper's mouths in this case) This article may have vaguely implied at the start that the Augan-dCS recorder sounds better than the rest of the equipment there (...of which there are 3 DSD systems and only 1 24/96 (not 24/192) PCM system. Way to go, beating 3 of your own kind and 1 weak representative from the other camp!) but the rest of the article is just a chain of mumbo jumbo about how hard it was to develop a proper DSD recording and mixing system. It never explicitly talked about the relative merits of DSD and PCM anywhere, certainly not from a technical point of view. Bit efficient, better time resolution, blah blah blah--you're making it all up.

Quote:

There is more to music than just bit resolution and that realisation is what is finally causing digital to pull away from vinyl based systems. One thing that a digital system should have over vinyl based systems is precision, however if your precision is focused on the wrong parameters, then you will have missed the mark by a long shot and I submit to you that SACD has made the right choice but choosing to preserve time resolution over its useful bandwidth.

By the Nyquist theorem, time resolution can be converted to frequency resolution. A system with higher time resolution must respond to higher frequencies.

Here are the frequency resolution figures for DVD-A and SACD:

Quote:

DVD-A 24/96: 0-48kHz, (0-96kHz for 24/192) 144dB dynamic range across the entire range SACD DSD: 0-100kHz (approx), 120dB dynamic range at lower frequencies, down to virtually nothing at higher frequencies (where everything is swamped by noise)

(Here's the source. It's a thread where I am actually floundering about for an explanation to DSD. If you doubt the claims above, you can post a question in that forum, where there are much more knowledgeable people than you or I)

Things to note:
1. SACD has lower dynamic range than 24/96 or 24/192 across the entire frequency spectrum it covers

2. SACD's frequency response range does *not* got up to 1.4MHz, as the 2.8MHz sampling rate may suggest. This is because of the noise shaping necessary to obtain acceptable dynamic range in the audible frequency range. Without noise shaping, the frequency response would be up to 1.4MHz, but the dynamic range would be only 6dB for the whole range!!

3. 24/192 can cover up to 96kHz with 144dB (i.e. 24bit resolution), while SACD would be close to 0dB at this point! The only advantage SACD has is that it responds to some frequencies beyond 96kHz, up to maybe a bit over 100kHz, but with the dynamic range closing to 0dB.

Translating this into time resolution: DVD-A can encode sonic events with cycling period > 1.04167e-5 seconds with 144dB of separation from background noise. SACD, in encoding a sonic event with cycling period close to 1.04167e-5 seconds, will give you an output that is barely distinguishable from the background noise!

Just to put this in perspective, our ear's time resolution only goes up to events with speeds up to a rate corresponding to ~20kHz (maybe higher for those with bat ears, but certainly not above 48kHz!), the frequency ceiling of our hearing. There are exceptional cases with transients and interaural time differences where we can respond to faster events, but any format that can record up to and beyond the frequency ceiling of our hearing can also encode these transients and ITDs perfectly.

Quote:

As I consistently said in this thread, I am yet to read anywhere that anybody has disputed the superiority of DSD time resolution over 192/24.

Here. *I* dispute the superiority of DSD time resolution over 192/24!

As outlined above I say it is worse

If you've read so much stuff that supports its superiority in this regard, maybe you can post some stuff to defend this. Heck, if you are so tech-savvy, maybe you can tell us what's wrong with that article somebody posted here that says SACD can't encode anything above 8000Hz correctly!

 

[Czilla9000]

Joe, it was the IAR-80 (www.iar-80.com) that says that DSD can't encode anything above 8kHz correctly. They point this out with vocals in particular. There observations also confirm my observations with DSD. When I listened to my DSD CD of the "1812 Overture" the first thing I noticed was that the vocal were off.


What I like about the IAR is they first listen to a product/system and see how it sounds. Then they go out and find a rational scientific explanation to explain why it sounds the way it sounds. The reason DSD sounds so smooth is it smear the high frequencies intentionally, because it can't really capture them to begin with.


However, it is worth pointing out that while the Compact Disk can capture high frequencies better than SACD, the CD still has trouble and benefits greatly from noise shaping. This is the true reason why CD players upsample and have digital filters. This is also why the dCs Purcell sounds so good. Most CD players can upsample to 24/192 anyway, but the Purcell has 9th order noise shaping as well.

It is also worth pointing out that back it the early days of the CD Philips came up with a method of producing budget CD players using a method they called (the envelope please) BITSTREAM! However, these budget bitstream players upsampled the signal 256 times to 11 MHz, as opposed to SACDs 64 times sampling. Keep in mind these 11 MHz bitstream players were developed by Philips only for the budget market. For the highend market they used 14/176 players. Wonder why.

(I would love to find one of those old budget bitstream players and sell it on eBay for thousands of dollars because it has "11 MHz upsampling!"

)


According to the IAR, DSD would have to upsample 32,000 times inorder to MATCH the high frequency performance of the CD.

That would be 141.120 MEGAHERTZ!

 

[Joe Bloggs]

Czilla9000,

I don't believe this applies to SACD as it stands now. But it will be up to theaudiohobby with his l33t technical knowledge to explain why this is so

 

[Czilla9000]

Quote:

I don't believe this applies to SACD as it stands now

Where did you hear this?



Also, let us treat each other with respect.

 

[theaudiohobby]

Joe and czilla9000,

read the papers before we discuss further at the rate you guys are going my response will probably be good enough for an AES paper.

suffice to say that the dcs paper does say this

Quote:

detailed comparisons not yet performed on enough systems, but well liked by (classical) artists after sessions · no “busy signal” break up · very good separation of reverberation and room acoustics · no observations on bass so far · strong stereo image formation, no observations on width so far

The philips response also says this

Quote:

10. The impulse responses of 4 different systems in a multichannel configuration are depicted: a 48 kHz system, with a bandwidth of 20 kHz (that is, 8 kHz transition bandwidth is allowed for anti-aliasing filtering), a 96 kHz system with 35 kHz bandwidth (26 kHz transition bandwidth), a 192 kHz system with 75 kHz bandwidth (42 kHz transition bandwidth) and an SACD system with 95 kHz bandwidth. Though none of the systems reproduces the input exactly, the DSD systems shows the least artifacts. Clearly, the 48 kHz system has great difficulty in reproducing the click; due to the steep filtering it starts ringing at a -30 dB level approximately 1 ms before the click, which is very audible. Also at the higher sampling frequencies, the ringing phenomenon cannot be removed, though it is reduced significantly. Only the DSD system is very effective in suppressing the ringing effect, due to very slow filtering above 95 kHz. The price to pay for this is the increase in noise floor with respect to the other systems; however, as the noise floor contains only high frequency components which are uncorrelated with the audio, they are not perceptible.

and that correlates exactly with the main decenting paper I gave you that says

Quote:

An alternative bitstream converter has been reported that uses a combination of a linear quantizer with dither to guarantee linearity, together with 4th-order noise shaping. Conversion to a 1-bit code (typically from a 4-bit code) is then performed by an open-loop, optimal code conversion table which minimises spectral modulation. Potentially the system produces high resolution with no low-level correlated distortion. However, the bit rate is again very much greater than PCM, and although solving the problems of correlated and idle-channel distortion it is too bit inefficient.

and philips answer to their objection is this

Quote:

Other issues which often appear to be confusing, are data rates in connection to the bandwidth claimed by SACD. The SACD format comprises (apart from its red-book conforming CD layer) two different music streams: a stereo 2 channel stream, and a surround 6 channel stream. Hence, an SACD contains 8 channels of high-quality audio. Because all channels are 2.8 MHz sample rate, 1-bit signals, the total data rate equals 2.8 Mbyte/s (or 22.6 Mbit/s). On these signals, lossless coding is applied. This lossless coding scheme is specifically developed for coding 1-bit signals. From experience of over 100 recordings, the average coding gain is roughly 2.4 - 2.5 for pop recordings, and 2.6 - 2.7 for classical recordings. This corresponds to a data rate per channel of about 1.1-1.2 Mbit/s. This indicates that on average 70 minutes of a DSD signal can be recorded on an SACD in the 8-channel format. For 6 channels, this amounts to roughly 95 minutes. Also, the high sampling rate of DSD allows for the use of filters with slow roll-off. We can compare this to DVD-A. The DVD-A format that gets closest to the SACD characteristics is DVD-A at 192 kHz, 20 bit, which reaches the same dynamic range, but is either of lower bandwidth than SACD if sloppy anti-aliasing filters are used, or has the same bandwidth using steep filters. Using a compression factor of 2 the data rate amounts to 1.9 Mbit/s, which is almost twice as much as the data rate for DSD. Hence, even if only six channels are used on the optical disk (compared to 6+2 on SACD), only 55 minutes of music can be stored - much less then the 74 minutes that we are accustomed to from CD.

I will return with some detailed answer to objections when the need arises. but as it is stands, it is easy to see here that SACD has a few aces up its sleeve that DVDA as it stands simply cannot match.

 

[kartik]

I am amazed by the level of this discussion and I think this thread should be preserved for the sake of future arguments between the technical and audio issues surrounding these formats. I wonder if this thread should be made into a sticky as this issue is going to arise again.. ...soon!

 

[Joe Bloggs]

theaudiohobby,

care to show us where you got those philips responses from?

Re: the dCS paper--for the LAST TIME!!!

DSD: no "busy signal" break up
24/192: no "busy signal" break up--excellent quality

DSD: very good separation of reverberation and room acoustics
24/192: very good separation of reverberation/room acoustics from instrument output

All seeing head to head up to this point, right?

24/192: but - bass can appear light and slightly out of time
DSD: no observations on bass so far

24/192: and - stereo image can be strong but widened (1.5 times?)
DSD: strong stereo image formation, no observations on width so far

How is a 'no observation' on the side of DSD supposed to be used to compare with any other format, whether favourably or unfavourably???

24/192 has 96kHz bandwidth.

They can use 'very slow filtering above 95kHz' with DSD to achieve good results on a graph, but this is a trick that has no meaning in real life, because in real life you have to start filtering way before 95kHz, otherwise your equipment may start having distortions that affect audio frequencies below 20kHz. AND, the so called 'very audible' -30dB ringing 1ms before the click, I can assure you, is very *inaudible*, since the ringing frequency is at 24kHz for the 48kHz system. Pre-ringing at 1ms before the actual sonic event *could* be audible, *if* it is in an audible frequency. 24kHz is not an audible frequency. If 24kHz is debatable, 48kHz (for 24/96) certainly is not.

I'll have to look things up about the compression ratio and get back to you.

Czilla9000,

I treat you with all respect

 

[theaudiohobby]

Guys,

here is philips response and it was published in the following of the dissenting paper

 

[JaZZ]

I wouldn't rely too much on what Philips claimed. The advantage of DSD opposed to high-rez PCM in the time domain (ringing!) is of rather theoretical nature given that the really existing filter characteristics of SACD and DVD-A players are very similar (relatively flat roll-off above 20 up to a sharper one around 80 kHz, depending on the player, but not the format) -- and the ringing lies primarily in the filter's responsibility.

So in reality DSD has no advantage over DVD-A. But well, ringing at say 40 kHz or above most likely has very few impact on the audible frequency range, in contrast to a 21- or even 24-kHz ringing, where it most likely has -- due to the fact that resonances have a certain bandwidth and don't restrict themselves to the exact nominal frequency. I have seen 20-kHz tone bursts, e.g. (from a redbook DAC), which show an insane amount of (20-kHz!) ringing and are barely recognizable as bursts anymore but rather look like continuous sine waves. Such as these:

The same applies -- in a diminishing degree -- to frequencies down to maybe even 10 kHz. I'm sure that this is the cause for the «digital glare» often criticized with the redbook sound.

DSD has indeed problems with high-frequency resolution in the bit-depth domain, because of not enough bits per time available to reach just redbook standards... I guess http://www.iar-80.com/ is right with this: A sampling rate of 2.8 MHz enables not more than 280 samples per cycle at 10 kHz (corresponding to 8.13 bit) bit), 175 (7.45 bit) at 16 kHz and 140 at 20 kHz (7.13 bit) -- further reduced by the fact that only triangle waves could fully exhaust the whole bit depth. Now I know that noise shaping is said to increase resolution (or just the signal/noise ratio?), but I can't imagine how the missing resolution data could be conjured up.

Nevertheless, I can't reproduce the sonic shortcomings of SACD observed by Peter Moncrieff from IAR. Having not heard DVD-A though, I have no clue if it's even better, as «promised». What I can say is that SACD is a clear improvement compared to redbook and offers obviously much higher resolution. It even hasn't to be anything with a lot of high frequencies to detect it: a simple grand piano is enough to hear it: rather boring with CD, alive, transparent and full of flavor with SACD. 44.1 kHz and 16 bit aren't enough to my ears, but SACD is good enough for me at the moment. Its advantage over DVD-A: clearly more available titles. But it's still a very poor choice. Luckily there are some highlights even aside the mainstream, e.g. on the Songlines label.

My next source device will be a universal player. I'm curious about DVD-A, but do I really need a TV to play music then?

 

[theaudiohobby]

Joe,

why do you persist in insisting that the glass is half full, look at this from the dcs paper,

Quote:

Recording engineers, and many musicians – particularly in the classical area – are becoming aware that material recorded and edited using these higher sample rates has some attractive qualities. Current theory on how human hearing works has so far been unable to explain the basis for these qualities, but they are none the less easy to demonstrate.

Read another quote
Quote:

The noise can be reduced by low pass filtering from somewhere between (depending on the system) 30 kHz and 60 kHz, but then this filter becomes a part of the system. These hf reduction filters have not so far received much attention – DSD users so far prefer to find systems that are hf tolerant, and minimise the extra filtering required.

My current power amp is almost ten years old year and has bandwidth up 130kHz, so the requirement for low pass filtering at 30khz or even 60Khz s not necessary. Secondly sony has speakers with response up to 70Khz, there are already modified players on the market with filtering kicking in @100khz, so the issue of filtering way below 96khz is is mute but it is already being done provided the equipment can support it as the paper indicates and under these conditions it will deliver a response that is superior to DVDA.

Furthermore look at their observation at 24/192, it says

Quote:

very good separation of reverberation/room acoustics from instrument output · but - bass can appear light and slightly out of time · and - stereo image can be strong but widened (1.5 times?)

in order words, unnatural, now read the albeit limited observations on DSD
Quote:

detailed comparisons not yet performed on enough systems, but well liked by (classical) artists after sessions · no “busy signal” break up · very good separation of reverberation and room acoustics · no observations on bass so far · strong stereo image formation, no observations on width so far

Any initials demonstration vis a vis the analog tape were favourable. Remember the issue of superiority I mentioned is transient response, now read this remark very good separation of reverberation/room acoustics from instrument
output·but - bass can appear light and slightly out of time
, Now it is clear also that so far no untoward was observed in bass frequencies of DSD that warrants question marks.

Now read this statement
Quote:

The bandwidth of 192 kS/s far exceeds the normal bandwidth attributed to human hearing, even using gentle roll off filters, so it is hard to believe it is related to frequency response. The stereo image widening is a very strong effect, observed by virtually all listeners, when a comparison with the analogue source is available. The bass problem is commented on by experienced listeners, and shows up particularly well on multi-mic’d rock music – it can enable 192 kS/s material to be identified in absence of the analogue source material for comparison. The bass response of the ADC and DAC units in question is flat down to 1 Hz, so there is no question of it being due to bass roll off there.

Now compare that to this comment taken in it's entirety,

Quote:

When played back on a system with adequate hf tolerance, the results from DSD are very good. We have not so far done enough comparative listening tests (DSD vs PCM or DSD vs Analogue) to comment usefully on the differences. We have done some testing of different filters in the noise shaping loop in a DSD modulator. However, even these results are dependent on the monitoring system and useful comments cannot be made at this stage.

Even they cannot comment usefully, it clear that the bass problems were not observed.

I will continue this response later