Spectrum from SACDs-Very Interesting..and dissapointing

[kap'n krunch]

Before you think that I'm bashing the format, I still enjoy it.
I just think that DVD-As are THE REAL McCOY if you want real Hi resolution Audio!!!(NOT counting BluRay, which I still have to buy a player )

What Neil Wilkes had exposed in this forum regarding his findings on their limited frequency response made me aware , but now I have (damning) further proof.

I sold 2 of my SACDs for dough and made a backup at 96/24 from the Analog Out of my Sony SACD player to the MOTU 828 mkII.
I recorded the WHOLE disc at once, so the wav files are of the whole disc.

This SACD player states that the Analog outs go up to 100KHz. Maybe they do, but it's just noise!!

I had never bothered to see the spectrum..until now.

And , man , my jaw dropped (in a bad way)...

Here they are , for your enjoyment and further discussion-let it rip folks!

Blue Oyster Cult-Agents of Fortune- Left Front Channel: NOTHING , and I mean nothing, above 20-21 KHz...JUST PITCH BLACK!!!-At least on my LP rips, I get the resonances from the cart/stylus to sweeten the sound)..
No wonder I always though that this SACD was VERY dark sounding...




Now , here is Groove Armada's -Vertigo Left Front Channel:
This one , surprisingly ( because of the mostly sampled-castrated at 20 KHz- nature of the sounds used here) has a HIGHER freq response than the BOC one, but not by much...
Again , pitch black above 23-24 KHz. Interesting peaks that "want to but can't" over the threshold...

 

[Disclord]

That's the problem with DSD - it's massively nose-shaped to shift all the noise from the 1-bit modulator up above 20kHz and if there is any actual musical spectrum, it has only about a 1db dynamic range above 20kHz due to the DSD processing. That's why Sony and Philips abandoned the 'real' 1-bit 64x over-sampled DSD recording system and used a multi-bit delta-sigma system for mastering and recording - but then it still has to destroyed to be put on the SACD in the 1-bit DSD form.

 

[kap'n krunch]

That's the problem with DSD - it's massively nose-shaped to shift all the noise from the 1-bit modulator up above 20kHz and if there is any actual musical spectrum, it has only about a 1db dynamic range above 20kHz due to the DSD processing. That's why Sony and Philips abandoned the 'real' 1-bit 64x over-sampled DSD recording system and used a multi-bit delta-sigma system for mastering and recording - but then it still has to destroyed to be put on the SACD in the 1-bit DSD form.

VERY INTERESTING!!!
Thanks for sharing this info,

 

[Disclord]

VERY INTERESTING!!!
Thanks for sharing this info,

Here's a link to an AES paper on my site about why the 1-bit SACD DSD approach is a bad system and inherently imperfectible.
http://issuu.com/disclord/docs/sacd-is-bad?mode=window&viewMode=singlePage

 

[winopener]

The best comparison you can make today is with the BR disc of DSOTM: rip the quad (analog sourced) and the 5.1 (dsd sourced) and compare the spectrum. A true desolation for the dsd.

 

[Lokkerman]

This debate has gone on in the SSGG/Hub for quite a while. Truth is that SACD gives you; depedent of the output brick-wall filter - (A lot of folks like the sound of Saracen), the PCM equivalent of around 21 bits at 48khz (or less). So lesser quality PCM of 24/48 is usually superior to SACD. If you ever get chance compare the brilliant Hotel California (#5.1) by the Eagles SACD versus 24/96. There is a veil lifted by the DVD-A.

Interestingly some DVD/SACD players have an ability to enhance the sound of SACD - My Denon does. Neil was correct a long time ago along with Stan Lipshitz and a few others. SACD was made to be a (cheap) replacement for CD but in reality was not in the same league as DVD-A. And I have a large collection of SACDs at that.

 

[kap'n krunch]

Here's a link to an AES paper on my site about why the 1-bit SACD DSD approach is a bad system and inherently imperfectible.
http://issuu.com/disclord/docs/sacd-is-bad?mode=window&viewMode=singlePage

Whoa, you guys are Rocket Scientists!!!
I can understand normal equations and normal logarithms, but this goes way over my head!

I'll try to read it later and see what I can get from it.

Thanks again.

 

[Disclord]

This debate has gone on in the SSGG/Hub for quite a while. Truth is that SACD gives you; depedent of the output brick-wall filter - (A lot of folks like the sound of Saracen), the PCM equivalent of around 21 bits at 48khz (or less). So lesser quality PCM of 24/48 is usually superior to SACD. If you ever get chance compare the brilliant Hotel California (#5.1) by the Eagles SACD versus 24/96. There is a veil lifted by the DVD-A.

Interestingly some DVD/SACD players have an ability to enhance the sound of SACD - My Denon does. Neil was correct a long time ago along with Stan Lipshitz and a few others. SACD was made to be a (cheap) replacement for CD but in reality was not in the same league as DVD-A. And I have a large collection of SACDs at that.

Yep, Sony and Philips CD patent pool was expiring and they wanted to extend the CD patents, plus get away from paying Discovision Associates for their optical disc patents - they tried the same thing with their original MMCD audio/video format which "kind of" failed by being blended with Toshiba's SD format into the DVD format we now have - so Sony and Philips used the DVD format as the 'base' for SACD with a CD layer glommed on for new patentability along with the terrible 1-bit DSD format to get away from expired PCM patents, although 1-bit was already old news (Sony's first low-cost consumer DAT deck, the DTC-700, used a 1-bit 64x noise-shaping A/D converter for recording). But 1-bit DSD has so many problems that they quickly abandoned it for mastering (without letting the public know) because it was so bad - plus had so much noise in the ultrasonic range that almost all SACD players were required to have a 30 kHz filter available on their outputs so as not to damage equipment that couldn't handle high levels of ultrasonic junk. Nor did they tell people that basically any signal processing to the 1-bit DSD signal turns it into a multi-bit PCM signal, which then has to be destroyed back to 1-bit DSD for storage on the SACD disc. Philips Direct Stream Transfer lossless compression only worked because half the spectrum was noise with a 1db dynamic range that could be easily predicted and flattened, allowing the bitrate to 'fit' on a DVD. MLP had to 'work' much harder, using pre-buffering, etc., because there could be real information and wide dynamic range in DVD-Audio's ultrasonic spectrum.

It's a shame DVD-Audio was handled so badly and that it's SACD's that are still being issued - MLP and it's extension in Dolby True HD is a wonderful lossless format - as is DTS HD Master, although it's a little less efficient. At least Blu-ray gives us a chance to have true high resolution lossless PCM audio, packed or not - with massive storage capability - and it's capable of perfection over its full bandwidth.

 

[SoNic67]

Hello to everybody! I know that I am new here and I sould not start with a disagreament, but I think that it is worth to know that ALL the modern DAC's are basically a DSD, 1 bit designs. So, for them, converting DSD is more natural to be done, with less processing than the PCM format (that needs to be brought internally to 1 bit format). No need for brickwall, no need for oversampling, dithering...
As for the PCM 192kHz - the present DAC's have a drop in audio performance (distortion) when they go over 96kHz.

As for the studio converstions from PCM to DSD, they have a chance to be done 1000 times better than the conversion that is done in the 4-10$ DAC chip.
Sure, variations in mastering don't allow a direct comparation between DSD and PCM. Those "papers" don't tell the whole story, and there are not necesarily not-bias.

Personally I feel that SACD is closer of the analog reel-to-reel and LP than the PCM. And I use the 50kHz slow roll-off filter.

 

[Disclord]

Hello to everybody! I know that I am new here and I sould not start with a disagreament, but I think that it is worth to know that ALL the modern DAC's are basically a DSD, 1 bit designs. So, for them, converting DSD is more natural to be done, with less processing than the PCM format (that needs to be brought internally to 1 bit format). No need for brickwall, no need for oversampling, dithering...
As for the PCM 192kHz - the present DAC's have a drop in audio performance (distortion) when they go over 96kHz.

As for the studio converstions from PCM to DSD, they have a chance to be done 1000 times better than the conversion that is done in the 4-10$ DAC chip.
Sure, variations in mastering don't allow a direct comparation between DSD and PCM. Those "papers" don't tell the whole story, and there are not necesarily not-bias.

Personally I feel that SACD is closer of the analog reel-to-reel and LP than the PCM. And I use the 50kHz slow roll-off filter.

First of all, glad you are here to argue an alternate point of view - everyone agreeing with each other isn't too much fun always.

DSD needs MASSIVE amounts of noise shaping, i.e. feedback, to push the noise floor down below 20kHz and move the noise up above into the ultrasonic. And it still needs dithering. The processing of PCM can be designed to be totally benign - not so with DSD.

Modern Delta-Sigma, so-called "1-bit" designs are actually 2 or 3 bit designs that don't have the problems of the SACD 1-bit DSD system. That's why Sony and Philips abandoned true 1-bit DSD for "DSD-Wide" which a multi-bit design for recording - and pretty much all high quality modern delta-sigma D/A converters are 2 or 3 bits - a sort of "MASH" type converter (which was always called 1-bit by Matsushita but really wasn't). The best, most expensive and high-precision A/D and D/A converters simply don't use true 1-bit designs like DSD - the industry has moved on to something better.

As for the drop in audio performance for present DAC's at 192kHz, that's a product design issue, not an inherent flaw of PCM at 192kHz, unlike 1-bit DSD which IS inherently flawed and un-fixable. All 1-bit DSD has above 20+kHz is lots of high level noise with a 1 db dynamic range to 100kHz or higher.

 

[SoNic67]

They are still 1-2 bit, instead of 24. Sure, they use fancy words to describe the noise shaping, interpolation and FIR filters that are included to make possible translation of PCM to delta-sigma.
But in the eend, the PCM needs to be converted to delta-sigma. That is either done in the DAC (PCM distribution) or in the studio (DSD distribution). I think that better results can be achieved in a studio enviroment (higher quality tools).

Also... "high noise" in audio band? I know that the dynamic in 20-22kHz is around 120dB. Sure, is not the maximum that a modern DAC can achieve (some 130dB) but in my house I cannot have sounds with more than 100-110dB dynamic anyway (some 18-20 bit). DSD quality is equivalent to 20 bit at 96kHz sampling - in audio band mentioned above. I cannot hear past 30kHz anyway .
The filters needed for correct recostruction of PCM are critical and are not always done right (anlog is almost a lost art).

Can be DSD improved with a higher edition like DSD128 used in Korg? Probably on paper. Not so much relevant in real life.

Is DSD-Wide is easyer to work with in studio? Is PCM DXD 352.8 kHz better? Yes and yes. Not relevant for distribuition.

To say that is inherently flawed is too much IMO. Unfortunatelly everything moves to Bluray. And really bad for audio - HDMI connections. Encrypted in real time, mixed with video (that has priority). Not really a proper connection mechanism for audio.

 

[beached]

The highest frequency that a normal middle-aged adult can hear is only 12-14 kilohertz. Also, the hearing range for men worsens more quickly than the hearing range for women. This means that women will have the ability to hear notes of higher pitch than men of the same age do. The lowest frequency we can hear is around 12hz.
Then there are the subsonics and ultrasonics which theoretically can extend aural awareness (via conduction) beyond these human hearing threshold points.

My thoughts are that I am not going to get anywhere near hearing or feeling any thing above 20 khz so the issue is of no real importance to me. Just give me stuff in surround and with at least 24/48 resolution.

 

[Bob Squires Jr]

The highest frequency that a normal middle-aged adult can hear is only 12-14 kilohertz. Also, the hearing range for men worsens more quickly than the hearing range for women. This means that women will have the ability to hear notes of higher pitch than men of the same age do. The lowest frequency we can hear is around 12hz.
Then there are the subsonics and ultrasonics which theoretically can extend aural awareness (via conduction) beyond these human hearing threshold points.

My thoughts are that I am not going to get anywhere near hearing or feeling any thing above 20 khz so the issue is of no real importance to me. Just give me stuff in surround and with at least 24/48 resolution.

God I miss AM Radio!

 

[Disclord]

The highest frequency that a normal middle-aged adult can hear is only 12-14 kilohertz. Also, the hearing range for men worsens more quickly than the hearing range for women. This means that women will have the ability to hear notes of higher pitch than men of the same age do. The lowest frequency we can hear is around 12hz.
Then there are the subsonics and ultrasonics which theoretically can extend aural awareness (via conduction) beyond these human hearing threshold points.

My thoughts are that I am not going to get anywhere near hearing or feeling any thing above 20 khz so the issue is of no real importance to me. Just give me stuff in surround and with at least 24/48 resolution.

I'm 43 and I can still hear up to 20kHz in my right ear - in my left ear however, I'm down to about 16kHz (I can still hear the fly-back transformer in CRT TV's but nothing above that). When I was a teen I was treated for cancer and the Cisplatin chemo used made my hearing so painfully sensitive that I could hear dog whistles and stuff and they HURT. Actually, it wasn't so much that I could "hear" them, I'd just get pain in my ears when one was used or very high frequency's were present. Listening to FM radio was painful if the radio didn't have good suppression of the 19kHz pilot tone. Apparently, it had something to do with the Platinum in the chemo - my oncologist said he'd encountered that before, especially in asthmatics (which I'm not). It took a few years for my hearing to go back to "normal" and I've tried to protect it ever since, such as wearing double ear plugs at Madonna's Blonde Ambition concert in LA.

 

[davie]

I don't see the point of bashing a format that is still fairly widely used and can handle 5 channels of music. As a music lover, that is a great improvement over a CD. When it is converted back to analog in the native format, without conversion to PCM, it's as good as anything I've heard. I am speaking now from the point of view of the consumer.

 

[Disclord]

I don't see the point of bashing a format that is still fairly widely used and can handle 5 channels of music. As a music lover, that is a great improvement over a CD. When it is converted back to analog in the native format, without conversion to PCM, it's as good as anything I've heard. I am speaking now from the point of view of the consumer.

Does it hurt SACD's feelings? In all seriousness, there was widespread anti-DSD sentiment by the industry (in particular those with deep knowledge of how PCM and DSD worked and could back up their statements with mathematical proof) when SACD was developed due to the well known flaws of the 1-bit storage format it used (flaws that were known since the 70's). DVD-Audio's PCM storage was capable of literal technical perfection, whereas SACD never ever can be. Sony and Philips cut their losses on the whole 1-bit format and Blu-ray allows infinitely higher quality now than SACD ever can.

The 5.1 channels of SACD (which many, many do not use) IS an improvement over CD - sound quality, not so much - in level-matched blind tests, SACD and CD can't be distinguished to any better level than guessing.

 

[beached]

in level-matched blind tests, SACD and CD can't be distinguished to any better level than guessing.

Then ultimately, to be fair, I think the same could be said for a comparison between say, Steely Dan's - Gaucho. SACD vs DVD A, both played in stereo mode of course. 1 Bit DSD vs 24-bit/192 khz. My 55yr old ears could not pick the difference, however I hear an added dynamic,detail and decay (3D) from both formats in comparison to the RBCD. I can not pick 24/48 from 24/96 or 24/192 either. 16/44 though, even at its best is not quite cutting it in comparison to even 20-bit 44.1 kHz and higher. IMHO, the heard improvement is all mostly due to a higher 20/24 bit rate (Detail & Decay) and not so much the extra 4/52/148 kHz (Dynamics) above RBCD 44.1 kHz. A bit more "punch" & "growl" from the extra kHz:howl, The older I get the harder it is to distinguish good DTS from SACD and 24/48kHz surround. Listen up while you can.

 

[ssully]

This debate has gone on in the SSGG/Hub for quite a while. Truth is that SACD gives you; depedent of the output brick-wall filter - (A lot of folks like the sound of Saracen), the PCM equivalent of around 21 bits at 48khz (or less). So lesser quality PCM of 24/48 is usually superior to SACD. If you ever get chance compare the brilliant Hotel California (#5.1) by the Eagles SACD versus 24/96. There is a veil lifted by the DVD-A.

I would be very, very careful to match levels before attempting to make such a comparison, especially if it's supposed to reveal a difference at 48 kHz between 24 and 21 bits (which I suspect no one who isn't a bat can tell apart).

Interestingly some DVD/SACD players have an ability to enhance the sound of SACD - My Denon does.

However it does that, it must first convert DSD to PCM.

Neil was correct a long time ago along with Stan Lipshitz and a few others. SACD was made to be a (cheap) replacement for CD but in reality was not in the same league as DVD-A. And I have a large collection of SACDs at that.

In technical terms, DSD has a some flaws, but the audible difference between it and high sample rate/bitdepth PCM is likely to be insignificant in a properly set up comparison. Results of the few such comparisons I know of in AES literature have been negative, or at best, for a very few listeners tested , inconclusive (i.e., seemingly 'positive' results achieved by a tiny minority of subjects were not retested to see whether they were flukes).

 

[ssully]

Here's a link to an AES paper on my site about why the 1-bit SACD DSD approach is a bad system and inherently imperfectible.
http://issuu.com/disclord/docs/sacd-is-bad?mode=window&viewMode=singlePage

...ISTR recall a 'reply' to that paper, and then a counter-reply from L&V....

Aha, Wikipedia helps me out here: http://en.wikipedia.org/wiki/Direct_Stream_Digital#DSD_vs._PCM

 

[ssully]

. But 1-bit DSD has so many problems that they quickly abandoned it for mastering (without letting the public know) because it was so bad - plus had so much noise in the ultrasonic range that almost all SACD players were required to have a 30 kHz filter available on their outputs so as not to damage equipment that couldn't handle high levels of ultrasonic junk.

IIRC Scarlet Book spec recommends either a 50 kHz or 100kHz lowpass output filter, the choice to be made by the player manufacturer.

Nor did they tell people that basically any signal processing to the 1-bit DSD signal turns it into a multi-bit PCM signal, which then has to be destroyed back to 1-bit DSD for storage on the SACD disc. Philips Direct Stream Transfer lossless compression only worked because half the spectrum was noise with a 1db dynamic range that could be easily predicted and flattened, allowing the bitrate to 'fit' on a DVD. MLP had to 'work' much harder, using pre-buffering, etc., because there could be real information and wide dynamic range in DVD-Audio's ultrasonic spectrum.

Some of the shenanigans pulled by Pyramix etc in their marketing re: DSD, was evaluated critically here:

http://web.archive.org/web/20071011...tre.org/surround2002/technology/page_07.shtml

That said, the effective dynamic range within the *audible* spectrum is quite good for DSD -- bit depth is circa 20 bits, or about 120dB. And excessive concern about missing information in the octaves above 22 kHz is misplaced, IMO (except for one instance: if one intends to correct for tape flutter, using something like Plangent's DSP...in that case the tape bias signal must be captured). The main flaw of DSD for me is that SACD is pointless. PCM as a recording, production, and delivery format can do everything important that DSD can do, and do it as well or better.