DVD-Audiobahn New DVD-Audio music releases and talk about the latest players.

---

[ Follow Ups ] Thread: [ Display  All  Email ] [ DVD-Audiobahn ]

---

[ Alert Moderator ]

---

Some more answers

Posted by Christine Tham (R) on June 14, 2004 at 14:37:36

In Reply to: Some answers posted by Frank.. on June 14, 2004 at 13:39:57:

Thanks Frank for supplying some of the answers. They are correct.

Here are rest of answers:

*** Why are the maximum and minimum sample values different for each of the formats? ***

Because they are different recordings - can't guarantee hitting the record button at exactly the same time etc and samples not quite at same positions in the music.

*** I see the clipping on the charts, but not in the tables (actually, the normalized DSD gets clipped in one sample?) ***

The tables just count how many samples hit 0dB FS (that's why it says "Possible clipping"). since i set the gain such that the entire recording is under 0dB FS, there should be no instances of peaks hitting 0dB FS (except for normalized DSD where it's been normalized so that 1 sample is at 0dB FS).

*** When doing the statistics, had you already filtered the sample as described in the end, or was it before? I'm not sure this makes a difference, but... again, curious. ***

Before. By the way, the link is still to the old article - these are not new results. I didn't bother posting the new results as they are so similar.

*** You also show the clipping in the DVD-Audio, directly related to the original clipping on the CD. ***

the clipping is not related to the CD. the CD clips in hundreds of places, the DVD-A only in a few.

*** So if you have no clipping in the DSD, it can mean many things: the transfer from analog tapes has been done more carefully for the DSD , or DSD handles dynamics better than PCM. It may also mean that clipping has been avoided by using a very good limiter... ***

correct. DSD also handles overload differently than PCM - i don't understand why but there are a few AES papers about it. i'm sure ted smith can explain. bottom line is it doesn't "clip" as badly as PCM - although it's best never to overload it as it does deteriorate the sound.

the AES paper i refered to would suggest it's best not to exceed -3dB FS for PCM to avoid any possibility of distortion when processing the analog reconstructed waveform. actually, to be conservative, -6dB FS, to allow for 6dB of "headroom" - but it would be very rare for the analog reconstructed signal to get this high.

*** Maybe I didn't understand a few things, but: in my little mind, if the difference between minima and maxima is greater over a sample, you have "more" dynamics, and RMS power measures that. ***

the difference between minimum and maximum RMS power is an indication of dynamic range. the SACD will always do poorly compared to the DVD-A/CD in this measurement because of ultrasonic noise.

Actually, in my new recordings (done on the DVD-2200) the CD recording has the highest dynamic range (around 89dB!!!) - this is because of lack of ultrasonic noise in the CD. it doesn't mean that CD is better than SACD or DVD-A.

What i refer to as "relative dynamics" can be roughly measured by difference between maximum and average RMS values. as you can see, SACD is higher than DVD-A which in turn is higher than CD on this measurement. it can also be measured by comparing the difference between any two arbitrary points on the waveform across the recordings.

*** It takes many operations in your paper to demonstrate higher dynamics: first normalizing, then computing an average, and then comparing the maximum to that average... when the original difference is actually higher for the DVD-A and CD than for SACD? ***

No, the relative dynamics differences can be observed simply by eye on the waveform - no computation is required. The computation is to measure the extent of the difference, but the differences have always been large enough to be visible to the eye even at screen resolution.

*** A contrario, if you have higher and faster peaks in the DSD version, and if this is consistent throughout the sample, then shouldn't you have a greater difference between minima and maxima in the statistics? ***

No, because of DSD ultrasonic noise, as I've mentioned. that's why it's better to look at difference between maximum and average RMS power, not maximum and minimum.

*** BTW: I still wonder about whether I should get rid of all that noise when I record SACDs, did you notice any difference on your system before and after you used the filter? ***

If you don't intend any post processing on the recording, then you don't need to filter. If you doing some heavy crunching on it (ie. you are doing a "remix") then filtering is a good idea.

DSD's weakness is that it is really only suitable for end delivery (i.e. listening to music), the ultrasonic noise complicates the editing process. To me, DSD is the equivalent of a vinyl "direct to disc" recording. Get it right the first time, and try and avoid any post processing. Or use it in the final mastering stage only - if you edit extensively, do it in PCM (heresy to the SACD fanatics, i know!).

This post is made possible by the generous support of people like you and our sponsors:

Signature Sound

[ Signature Sound Lounge ]

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---