Thread: Groundbreaking DXD recording for SACD

Posts: 36
Page: prev 1 2 3 4 next

Post by DAD-Digital Audio De December 23, 2006 (21 of 36)
Immortal Nystedt was recorded in PCM. I am sure Lindberg has a lot to say about how it was done.
I do not know if the Mahler 2nd was recorded in DXD.
I could perhaps be the case if it was recorded by Andrzej Sasin.

Dear Windsurfer if you have it could you check. I am very keen to know every time something has been recorded or even only edited in DXD.

Many thanks

Peter

Post by Lindberg December 23, 2006 (22 of 36)
DAD-Digital Audio De said:

Immortal Nystedt was recorded in PCM. I am sure Lindberg has a lot to say about how it was done.

Five omnidirectional microphones directly to their respective speaker in MCH. Recorded in 48k/24B as this was prior to our transition into DXD.

Post by gfresh October 18, 2007 (23 of 36)
An old thread, but I was really interested in hearing the DXD technology in action. First, the performance and repetiore were both phenomenal! Even if the sound on this one sucked I would still probobly want to own it. The dynamics are awesome and Marianne Thorsen's violin is incredible.

So for the DXD compared to DSD I put my ears to the test. I have worked with analog tape, DSD and all PCM bit depth and sample rates. I was actually not suprised by what I encountered. This is sonically among the best SACDS I have heard, there is still a tradeoff. The clarity, detail, micro and macrodynamics and high frequency response (the smooth squeeks on the solo violin) are better than any other DSD or analogue sourced SACD I have heard!

However, I felt the realism and spatial imaging fell between 24/192 and 64fsDSD and like 24/192 the soundstage was trying hard, and fairly impressive but not quite satisfying. When I listen to a pure DSD recording the highs may be overly smoothed, dynamic resolution ok, but it always feels right, especially in the low frequencies. I think this has to do with more accurate timing and impulse response of DSD as result of bandwidth and lack of anti-alias filter. DSD recordings just have a natural fullness and ease that you can feel.

The DXD is getting close, but it is still a little ackward because it is still not quite there. The soundstage is subtley flatter, the bass fills the room less and the upper mids can be stiff in a very typical PCM kind of way.

I think this corresponds with the impulse graph, DXD gets 88% of the pulse and is about half as fast as DSD in the time domain. It doesn't look totally up to spec on the graph, and it kind of sounds like it looks. It would be nice to assume that 88% of the pulse is close enough that the human ear can't hear the difference, but the recording still sounds PCM to me. Amazing PCM, but not quite there. The DXD sounds better than straight DSD in terms of detail and accuracy, but the fullness and resonance is not on par which just makes me wanting more. It just doesn't sit quite right.

However, I am very excited for this new high resolution technology. I hope once the DXD technology gets going that the sample rate gets moved up to say 500khz so then we really have a minimal difference in soundstage from DSD with the benefit of full 24bit resolution and excellent frequency response. That would seem to be a good medium between the current standard and "perfect" filtering at 700khz which would require more and more noise shaping, like DSD. I would also wager the difference in that last 200khz would be almost imperceptable.

Well, that or we could go to DSD128.

Post by dvda-sacd October 19, 2007 (24 of 36)
Well, we could go to DSD 256 fs, without noise shaping. ;-)

DXD is an excellent format for digital production.... but I think the search for perfection must go beyond DXD.

Cheers!

Post by Myrtone October 19, 2007 (25 of 36)
Well, I know what DSD is (1 bit, oversampled and noise shaped), but how does DXD work, I haven't seen any diagrams, I know that PCM is like a stepped waveform and DSD like an odd square wave, but I haven't found any oscilloscopic representation of DXD.

Post by The Seventh Taylor October 20, 2007 (26 of 36)
As far as I'm aware DXD is basically 'just' PCM with 352.8 kHz sampling rate and 24 bit resolution.

/faq#audio12

Post by Myrtone October 24, 2007 (27 of 36)
So, the sampling rate is lower than DSD? The FAQ doesn't call it PCM, but PCM like. http://www.digitalaudio.dk/technical_papers/axion/dxd%20Resolution%20v3.5.pdf doesn't suggest it is the same as PCM.

Post by Karlosak October 24, 2007 (28 of 36)
Yes, the sampling rate is lower (352.8kHz) but at 24 bit resolution. You're right that the paper calls it only a PCM-like signal, but it is really nothing different.

Post by Myrtone October 25, 2007 (29 of 36)
But if you look at the frequency response, it looks not quite like PCM (which is in theory flat under half the sampling rate), you can see the rolloff near the top. It actually uses a 32 bit floating point.

Post by Karlosak October 25, 2007 (30 of 36)
The "flatness" of PCM at lower sampling rates is due to the steep filters used. They are needed to preserve as much of high frequencies as possible, unfortunately at the expense of some ringing introduced.

As we move up to higher frequencies, the need for such steep filters lessens, we can use a more relaxed attenuation rate. The fact that the signal is not fully attenuated at half the Nyquist frequency doesn't matter much, since the signal level is very low at these high frequencies and thus no aliasing artifacts threaten.

As you can see in Diagram 2 in the DXD paper, even PCM 24/192 (green line) starts to fall down at 60kHz with full attenuation at around 150kHz.

Page: prev 1 2 3 4 next

Closed