To ‘Improve’ DSD by Converting it to PCM (1 Viewer)

[Ed St. Clair]

High Fidelity Review
SACD News Story:


Audio Aero To ‘Improve’ DSD by Converting it to PCM.

Audio Aero’s high-end SACD player takes the unusual step of deliberately converting DSD to PCM in order to improve the standard of playback.

The Prestige A/V player, available in late 2002, is a combination DVD-Video and SACD machine which implements Audio Aero’s S.T.A.R.S.® process, a proprietary 32/192 re-sampling technology. It is used to enhance the playback of DVD-Video, CD, CD-R and CD-RW, and can also be applied to signals routed via the player’s own digital inputs (front left and right channels only).

Uniquely, for SACD playback, the Prestige A/V player uses 24bit 192kHz PCM digital to analogue conversion with what Audio Aero describe as a “DSD/PCM bridge”.

Intrigued by this, I asked Jean-Paul Combelles, technical and commercial manager for Audio Aero just what is meant by “DSD/PCM bridge”?

“The DSD to PCM bridge is a DSD to PCM converter that affords a 9dB improvement to the SACD signal-to-noise ratio by reducing the high frequency noise present due to the 2.82MHz sampling frequency. This will also allow enhanced functionality from the analogue stages,” explained Jean-Paul.

This statement actually confirms what many DSD sceptics have been citing as a negative aspect of SACD for some time, so I wanted to clarify just what Jean-Paul meant. I asked him to confirm that all DSD (SACD) sources are converted to PCM in the player, where a filter removes some of the ultrasonic noise present and thereby improves the signal-to-noise ratio by 9dB. I also asked that if this really were the case, whether he thought that SACD proponents would object to DSD being converted to PCM prior to D/A conversion? A pure DSD signal path appears to be the panacea of many SACD listeners.

Jean-Paul reiterated his earlier response: “…suppressing the ultrasonic noise will allow better functioning of the following stages (DAC and analogue stage). By reducing this noise, the residual signal we create will be boosted by 9dB using our enhancement techniques.” Unfortunately any more technical detail is considered confidential.

 

[Lee Scoggins]

A pure DSD signal path appears to be the panacea of many SACD listeners.

Yes, indeed. But don't take my word for it: listen to the pure DSD recordings out there. IMHO, they are a big step above regular Super Audio recordings.

 

[Kevin C Brown]

Next thing people will be saying, is that you can make lps sound better with a PCM conversion.

 

[John Kotches]

Lee,

If you read the words carefully, you will not see any statement that the ultrasonic noise is audible, only that they have increased inband performance by 9dB with the inclusion of additional filters.

How can you say it's a stupid idea when you haven't heard the player yet? Personally, I'm interested to hear what it sounds like -- more for curiosity's sake than anything else. Given the cost of their Redbook players this'll be deeper than my pockets will be able to take in the forseeable future.

Besides, I'm sure that they aren't the only company that is (or will be) doing a DSD to PCM conversion for SACD playback.

Regards,

 

[Ryan Spaight]

Um, doesn't the word "ultrasonic" imply "inaudible"? So in essence you're paying a bunch of money for this gizmo to eliminate noise you can't hear?
Or have we entered the Audiophile Doublespeak Zone?

Next thing people will be saying, is that you can make lps sound better with a PCM conversion.

Sure, it's called a CD remaster!
Ryan

 

[Justin Lane]

Um, doesn't the word "ultrasonic" imply "inaudible"? So in essence you're paying a bunch of money for this gizmo to eliminate noise you can't hear?

Isn't increased (ultrasonic) frequency response one of the selling points of SACD and DVD-A? We are paying a premium for frequencies we can't hear in recordings, so we might as well pay to remove them as well
J

 

[Lee Scoggins]

Isn't increased (ultrasonic) frequency response one of the selling points of SACD and DVD-A?

Justin,
Yes it is, but some like John believe you can hear the effects of this noise in the audible band. As a working recording engineer I have to tell you my team has never heard it.
On the contrary, we feel that only DSD gets the sound of a cymbal right or the impact on individual notes (the transients to be exact - attack, sustain, decay) perfect.

 

[Lee Scoggins]

Next thing people will be saying, is that you can make lps sound better with a PCM conversion.

Kevin,
You are close, there is a company that is doing DSD to LP conversions for better LP performance.

 

[KeithH]

A bit off-topic, but I have to wonder why a company like Audio Aero would include DVD-Video playback in this component. Interesting. Does a high-end company like Audio Aero need to include DVD playback to bolster sales of this player? Like they are going to sell a lot of them anyway. My point is that inclusion of DVD playback could hinder the sound quality of this player. Why complicate matters by including DVD circuitry?

I would very much like to hear this player, but the idea of converting from DSD to PCM bothers me.

 

[KeithH]

For $hits and giggles, Audio Aero should make this player upsample CDs to DSD. They could market it as a real crossover component.

 

[John Kotches]

Lee,

It's hard to comment on products that aren't actually in the marketplace, or on recordings made with those same products. If/when some recordings are made on these products, then we'll be able to discuss their merits.

Further, these products have a huge entrenched base to sell against, so I doubt that they'll have a major impact in the forseeable future on any appreciable scale.

In other words, time will tell on all of this.

The Audio Aero post is not critical of SACD in any way shape or form, my commentary was strictly limited to your statement that it's the stupidest audio idea you've heard of, even though you haven't heard the actual product.

Regards,

 

[John Kotches]

Keith,

You can already buy material that "upsamples CD to DSD". Look for Sony SACDs with digital masters originally recorded in the Mid 1980s to the late 1990s.

Regards,

 

[Ken_McAlinden]

I always was under the impression that part of the elegance of DSD was the ability to implement a high quality digital to analog converter much more simply than for PCM. The process of repurposing DSD to PCM so that you can apply extra DSP to the signal would seem to run counter to this.
I will assume a somewhat skeptical "I'll believe it when I hear it" position on this issue until further notice.
Regards,

 

[KeithH]

John, I assume you are referring to Sony's SBM process. I have a number of CDs and SACDs remastered using the SBM technology. In any event, I was referring to having the player "upsample" the PCM bitstream off a CD to DSD before conversion to analog, the opposite of what it seems the Audio Aero player will do with SACDs.

 

[Lee Scoggins]

I always was under the impression that part of the elegance of DSD was the ability to implement a high quality digital to analog converter much more simply than for PCM. The process of repurposing DSD to PCM so that you can apply extra DSP to the signal would seem to run counter to this.

Ken,

This is true. That is why I said it was the stupidest idea I had heard in some time.

In my opinion, I feel DSD is a step change better than PCM. Part of the reason involves faster sampling rate and part involves fewer circuits like complex brickwall filters. To add more PCM at the end of a DSD circuit path seems to defeat completely the original purpose of DSD.

There are also ways to decrease the ultrasonic noise that is much discussed. You can apply dither to reach noise levels of -160 to -180 db as everyone agreed at 2 AES conferences or so ago.

John,

There are pure DSD editing stations already being used. I know because I have seen them. You seem more upset by the fact that this takes away a convenient argument of yours that PCM occurs in DSD recording. But there are always crossover technologies that occur when a new format or technology arrives on the scene.

As we can tell from the new Mahler recordings by SFS, the results of pure DSD are well worth the wait and the highest expression of what this great format can do.

 

[Ed St. Clair]

20Hz to 20kHz is the 'recognized' boundaries of human hearing.
Of coarse there are always exceptions, with special cases of hearing well beyond 20kHz. None the less, only one out of a hundred of us, and then only when we are young, can hear 20Hz to 20kHz.
This was the 'theory' of the CD red book of a "brickwall" at 22.05kHz. Thinking 'we' can't hear anything over 20kHz.
'We' generally can't.
However, 'we' sense frequency's both below & above the 20/20 threshold.
Hence the failure of 16/44.1 to 'capture' the true playback of music.
It's going to be interesting too see if this 'rounding off' of DSD will enthrall listeners.

 

[Ken_McAlinden]

There's theory and then there's reality. While 22.05 KHz is the theoretical Nyquist Rate "brick wall" for sampling at 44.1KHz, you have to contend with the fact that your real ability to reproduce the sampled frequencies actually begins to deteriorate as you approach this frequency. Oersampling helps to eliminate this concern.

In addition, before sampling, analog anti-aliasing filters are applied to the signal to make sure that no spurious signal over half the sampling rate is present to be confounded with the lower frequencies. Oversampling allows one to shift the cut-off frequencies of the analog anti-aliasing filter well outside of the audible range. The oversampled signal can later be filtered digitally before downconverting to a lower frequency rate.

Regards,

 

[Lee Scoggins]

Oersampling helps to eliminate this concern.

Yes, but there are complexities involved to produce the brickwall filter as none are perfect.

The best way is to use DSD to eliminate the need for the filter altogether.

 

[Ken_McAlinden]

I was responding to a specific statement about the PCM 22.05KHz brickwall, but since you mentioned it, A signal with frequency content that DSD could not keep up with would cause problems as well, so signals must be filtered before sampling. Both oversampling PCM and the high sample rate of DSD allow the cut-off frequency for such filtering to be pushed way out of the audible frequency range.

Regards,

 

[Lee Scoggins]

Both oversampling PCM and the high sample rate of DSD allow the cut-off frequency for such filtering to be pushed way out of the audible frequency range.

Agreed.