Spectral Analysis : DVD-Audio Vs. CDDA, Making sense of the pictures? Options

[encosion]

First of all, I'm clearly new to both spectral analysis and DVD-Audio... And it's possible I haven't ripped the DVD-Audio disc properly either... Take a look at these images and let me know what you think:

The source material is from Björk's Medúlla, which is a heavily processed vocal only album:

CDDA 2.0 16bit 44.1khz


DVD-Audio 2.0 24bit 96khz


DVD-Audio 5.1 24bit 96khz






Have I ripped it correctly?

[Rotareneg]

On the "DVD-Audio 2.0 24bit 96khz" one it looks like the only audio above 22050 is from clipping distortion and a bit of aliasing. On the "DVD-Audio 5.1 24bit 96khz" ones everything above 22050 is just aliasing.

[encosion]

Does this means its actually 48khz? (I tried ripping a track at 48000 but the WAV played back half speed in Audition)

So they've used "quarter resolution" (22khz), upsampled to "half resolution" (48khz), to save space on the disc? It's a DL disc with DVD-Audio in 2.0 and 5.1 plus the usual DD5.1, DTS 5.1, PCM 2.0.... AND a 2Gb documentary VOB file too... Which means they had to drop the resolution to fit everything on right?

[dreamliner77]

**EDIT**

Geez, I don't read every thread for a few days and look what I miss.

This post has been edited by dreamliner77: Jul 10 2005, 02:33

[Mike Giacomelli]

The DVD-A rips have the same frequency range as the CD. So either you ripped it wrong, or they ripped you off

[encosion]

Considering they stuffed a helluva lot into 7Gb ("DVD-Audio in 2.0 and 5.1 plus the usual DD5.1, DTS 5.1, PCM 2.0.... AND a 2Gb documentary VOB file too...") it's probably not surprising that I've been "ripped off"... I'll not complain though, the multichannel 24bit surround still sounds fantastic, even if only in a resolution slightly superior to CDDA!

I ripped another DVD-A disc too... Here's a spectral snapshot of one of the DVD-A 5.1 channels:



This is from an album that was originally released in 1998 on CDDA, remastered in 5.1 just last year - it actually won the award for Best Multichannel Reissue at the Surround Music Awards in the US in August last year.

Looking any better?

[encosion]

Here's the frequency analysis of one of the channels of the Bjork DVD-A 5.1:


And here it is for "Refused" (The other DVD-A disc I ripped):

[Gabriel]

Considering they stuffed a helluva lot into 7Gb ("DVD-Audio in 2.0 and 5.1 plus the usual DD5.1, DTS 5.1, PCM 2.0.... AND a 2Gb documentary VOB file too...") it's probably not surprising that I've been "ripped off"... I'll not complain though, the multichannel 24bit surround still sounds fantastic, even if only in a resolution slightly superior to CDDA!

I ripped another DVD-A disc too... Here's a spectral snapshot of one of the DVD-A 5.1 channels:



This is from an album that was originally released in 1998 on CDDA, remastered in 5.1 just last year - it actually won the award for Best Multichannel Reissue at the Surround Music Awards in the US in August last year.

Looking any better?

Incredible! It features the TV pilot line at about 16kHz, and this pilot line seems to be aliased to 31kHz.
At least visually it seems to not be so good.

[encosion]

Could you explain what a "TV pilot line" is? I can see what you're talking about, but what does it mean?

P.S. The track is actually called "Liberation Frequency", so it could be conceptual... Here's a snippet of the lyrics for kicks: "It's coming through the air for all of us to hear. Could it be the sounds of liberation or just the image of detention? We want the airwaves back ... We don't just want airtime we want all the time all of the time ... We want transmission for the people by the people. What frequency are you getting? Is it noise or sweet sweet music? On what frequency will liberation be? ..."

This post has been edited by encosion: Jul 10 2005, 16:37

[Frank Klemm]

Incredible! It features the TV pilot line at about 16kHz, and this pilot line seems to be aliased to 31kHz.
At least visually it seems to not be so good.

The source depends on the exact frequency.

15.625 kHz + 31.250 kHz: horizontal deflection + 2nd harmonic (CCIR or OIRT)
15.734 kHz + 31.468 kHz: horizontal deflection + 2nd harmonic (FCC)

15.625 kHz + 32.375 kHz: horizontal deflection + aliasing on 48 kHz
15.734 kHz + 32.266 kHz: horizontal deflection + aliasing on 48 kHz

15.625 kHz + 28.475 kHz: horizontal deflection + aliasing on 44.1 kHz
15.734 kHz + 28.336 kHz: horizontal deflection + aliasing on 44.1 kHz

31 kHz looks more like 2nd harmonic (of the sawtooth horizontal deflection).

EDIT: Visit the image again. There is a line at about 30,5 kHz and another one about 31,3 kHz. And then there are lines at 42 and 44 kHz.

But your're right: All above 20 kHz is noise or distortion, i.e. shit.
An the loudness race also starts on DVD-A.

This post has been edited by Frank Klemm: Jul 10 2005, 17:07

[CSMR]

How did you rip the DVD audio? That isn't possible with commercially available equipment apparently.

[encosion]

http://www.hydrogenaudio.org/forums/index....showtopic=34368

[SirGrey]

Could you explain what a "TV pilot line" is? I can see what you're talking about, but what does it mean?

In a TV broadcast video pilot frequency is inside audio signal (or in some very old cases it was an audio signal itself)...

Anyway, this record seems to contain some pilot line for broadcasting

[Mike Giacomelli]

Could you explain what a "TV pilot line" is? I can see what you're talking about, but what does it mean?

In a TV broadcast video pilot frequency is inside audio signal (or in some very old cases it was an audio signal itself)...

Anyway, this record seems to contain some pilot line for broadcasting

So the source for the DVD-A was actually a recording made for broadcast TV?

[encosion]

Ok... I can confirm that signal only appears in that track... How's my luck I picked that track!

[Cyaneyes]

That signal appears in many many many recordings. It can be caused by something as simple as a TV being on in the recording studio.

[Borisz]

Here's some more

http://www.evilboris.sonic-cult.net/dvda/

I put up analysis pics for the track "Dance of Death" from the self-titled Iron Maiden DVD Audio disc. The MLP versions were both 88200Hz, with the 6chn version being 24bit while the stereo mix was 16bit. The rest is from the DVD Video part of the disc, a stereo LPCM and two 6chn rips - AC3 and DTS. All of them being 48khz. The AC3 was 448kbps while the DTS was 1509 kbps.
For the 6chn rips, the channel order is left, right, center, LFE, rear left, rear right.

Unfortunetaly, I don't have the CD version of this album to compare
Does anyone have it for some quick comparisons?

[necropimp]

what are you using to get those images?

[Borisz]

Adobe Audition / Cool Edit Pro (same thing, Adobe bought and rebranded it)

[Pio2001]

Acoustic or electronic music don't expand very much above 20 kHz. It's not surprising to see the audiogram decreasing above this frequency.
Moreover the signal will be limited by microphone or recording amplifiers frequency response.

Incredible! It features the TV pilot line at about 16kHz, and this pilot line seems to be aliased to 31kHz.
At least visually it seems to not be so good.

In order to get a proper idea, we need the color scale of the audiogram. A TV pilot line recorded at -120 dB is completely harmless.
If the audiogram was more detailed in the 40-100 Hz zone, we would very likely see a 50 or 60 Hz component too, caused by the presence of electric plugs in the recording studio.

[Scipio]

In order to get a proper idea, we need the color scale of the audiogram. A TV pilot line recorded at -120 dB is completely harmless.

so how do you get to see the color scale of the audiogram ? is that what some of those specialised video scopes are for ?

is there a way to do it in software ?

[Pio2001]

We just need to know how much decibels are each color in the posted pictures.

For example,
Yellow = 0 db
Red = -10 dB
Purple = -20 dB
Blue = -30 dB
Black < -30 dB
Would be very annoying, baceause it would mean the the pilot tone would be louder than -30 dB.

On the other hand, if
Yellow = 0 dB
Red = -40 dB
Purple = -80 dB
Blue = -120 dB
Black <-120 dB

Then the pilot tone is inaudible, since it is around -80 dB.

[precisionist]

is there a way to do it in software ?

In cool edit, just select "spectral view"...

We just need to know how much decibels are each color in the posted pictures.

For example,
Yellow = 0 db
Red = -10 dB
Purple = -20 dB
Blue = -30 dB
Black < -30 dB
Would be very annoying, baceause it would mean the the pilot tone would be louder than -30 dB.

On the other hand, if
Yellow = 0 dB
Red = -40 dB
Purple = -80 dB
Blue = -120 dB
Black <-120 dB

Then the pilot tone is inaudible, since it is around -80 dB.

Nice explanation, Pio. I consider myself a skilled cool edit user, but afaik there's nowhere such a scaling. In the help it just says that blue is quiet and yellow is loud and such.
Anyway, I don't find spectral view very informative. To messure available frequency content, I'd rather use the frequency analysis window, not only because of the lacking colour scale.
@Borisz: I'd prefer much more to see the amplitude-time screenshots..

edit: Maybe I'll try to generate sine waves at known frequencies and amplitude so I can find out a rough relation between the colour and its loudness...

This post has been edited by precisionist: Jul 11 2005, 17:10

[necropimp]

here's some of The Collector by Nine Inch Nails complete with crappy jpeg compression

CD audio was resampled to 96khz to show the massive empty spot above 22.05KHz (this is where someone points out an easier way to show the same thing without resampling a 44.1KHz file to 96KHz making me feel dumb for a moment)

http://jim7.home.insightbb.com/random/collectorCD-DA.jpg
http://jim7.home.insightbb.com/random/collectorDVD-A.jpg

[Frank Klemm]

here's some of The Collector by Nine Inch Nails complete with crappy jpeg compression

CD audio was resampled to 96khz to show the massive empty spot above 22.05KHz (this is where someone points out an easier way to show the same thing without resampling a 44.1KHz file to 96KHz making me feel dumb for a moment)

http://jim7.home.insightbb.com/random/collectorCD-DA.jpg
http://jim7.home.insightbb.com/random/collectorDVD-A.jpg

Please an also a plot of the waveform, to see where are overloads possible.