DVD-A Mastering, Post your analysis Options

[Cyaneyes]

I debated adding this to one of the existing DVD-A threads, but they're getting a little congested.

With DVD-A now rippable, let's use this thread to analyze DVD-A mastering and, if possible, how that mastering compares to the CD version of the album. The primary difference will hopefully be that DVD-As are unaffected (or less affected) by the loudness race.

Here's a look at "Deadwing" by Porcupine Tree.

These are the first four songs of the album from the CD, wavegained to be the same perceived level as the DVD-A.



These are the same four songs from the 5.1 DVD-A, converted to stereo with fb2k. As you can see, quite a bit less compressed.



Here's a closeup of the first two peaks of the album from the CD. Note the clipping.



And the same two peaks from the converted 5.1 DVD-A. Goodbye compression!


Album gain, CD: -9.83 db
Album gain, 5.1 DVD-A converted to stereo: +1.08 db

I haven't ripped the stereo PCM layer of the DVD-A yet. We'll see if that's closer to the levels of the CD or the 5.1.

This post has been edited by Cyaneyes: Jul 15 2005, 17:03

[ChangFest]

I think the stereo track on Deadwing is most likely the same version as the CDDA version. You'd probably see the same clipping and squashing on that as you do on the CD version.

This post has been edited by ChangFest: Jul 17 2005, 17:50

[Cyaneyes]

You're right, it was. Boo.

No big loss, the 5.1 converted to stereo is great.

[Acid8000]

That squashing of the audio on the CD version looks horrible, yet is so familiar nowadays.

[Vanishing]

The same principle normally applies for SACD as well, here is a comparison of "Desert Rose" from Sting from the SACD of "Brand new day":



The version on top is from the CD-layer, the other version is a downmix of the 5.1 SACD-layer, both replaygained. The measured gains are:

CD: -6.51dB (Album), -7.41dB (Track)
SACD 5.1: -4.00dB (Album), -5.91dB (Track)

These are all relatively harmless gains, but the difference in quality really is very noticable.
Other examples I know of are Metallica S&M, where the AC3 sound from the DVD-Video is less compressed than the CD version, or the DVD-A of "Audio" by the Blue Man Group, which is, with an album-gain of +0.16dB, a really dynamic surround mix.
Like many others here I really hate over-compressed music and therefore enjoy listening to DVD-As and SACDs, not because of their higher sampling-rates or bit-depths, but because usually they're mastered not with overall loudness as the ultimate goal.

[rasth]

The version on top is from the CD-layer, the other version is a downmix of the 5.1 SACD-layer, both replaygained. The measured gains are:

How did you downmix the SACD layer? In analog or on your copmuter?

[Vanishing]

I recorded the different channels of the SACD (sperate analog outputs of the player into 8 channel soundcard), converted them to a multichannel wav-file and used foobar's downmix DSP (Convert 5.1 to stereo).

[precisionist]

Thank you so much, Cyaneyes.
the second screenshot in your first post (5.1 to stereo): Does it really have the same average level like the clipressed CD version ? Because it looks like a much higher average loudness than that screenshot.
Nevertheless, it has much more dynamics.

[Cyaneyes]

Thank you so much, Cyaneyes.
the second screenshot in your first post (5.1 to stereo): Does it really have the same average level like the clipressed CD version ? Because it looks like a much higher average loudness than that screenshot.
Nevertheless, it has much more dynamics.

Yes, they're wavegained to the same perceived loudness. Almost 11 db difference in replaygain values.

The 5.1 looks louder in that screenshot than it really is. If you zoomed in on it, you'd see a relatively low average level with peaks far above.

[precisionist]

That's absolutely great. Roughly +1 dB adjustment to match the wavegain reference level! How does foobar convert 5.1 channels to stereo wav, does it just add the single channels ? And still a positive gain..

[Cyaneyes]

I assume it's just mixing the single channels, yeah. Also, I did have to apply a preamp of -12 db with foobar so that the resulting stereo mix didn't clip. Adding the six channels together apparently makes the downmix louder.

With the -12db adjustment, the peak value of the album was around -2 db. I then normalized to 100% and the resulting album gain value was +1.08.

[Vanishing]

I don't know how the foobar downmix works, but if you're interested in the topic of downmixing (which is more complicated than you might think), I can recommend the following page:

http://www.hauptmikrofon.de/downmix.htm

Apparently effects like comb-filtering and broadening of the stereo-image can occur depnding on the microphone-technique used for recording.

[RockFan]

Like many others here I really hate over-compressed music and therefore enjoy listening to DVD-As and SACDs, not because of their higher sampling-rates or bit-depths, but because usually they're mastered not with overall loudness as the ultimate goal.

You don't suppose 'they' were counting on exactly this when they started as-good-as ordering engineers to deliberately upf*ck CD releases?

Nah - they wouldn't dream of trying to force people to buy a new format that allowed the kind of extortionate mark-ups they alway appeared to regard as their God-given entltlement.

Unfortunately none of the high-rez formats are ever going to be diverse enough in releases to satisfy electic tastes, so they've ruined (for the most part) am otherwise perfectly adequate format for nothing. Stupid bastartds

R.

[Vanishing]

... so they've ruined (for the most part) am otherwise perfectly adequate format for nothing.

Who do you mean by "they"? If you mean the producers/record companies I would agree.
If you mean the engineers, I think it's not always clear who is responsible for over-compression or clipping in the final music. It could happen as early as during the recording of the individual instruments, e.g. recording a voice with heavy compression or tape-saturation. In that case the recording engineer would be responsible. Or it could happen during mix-down, when too much compression or even limiting is applied to individual tracks or subgroups. That would be the mixing-engineer.
In both cases, even a good mastering engineer can't undo the damage and often gets blamed.

Of course also the mastering engineer can mess up a mix, but normally he has more sophisticated equipment and techniques and can get a recording louder with less of the horrible-sounding byproducts like clipping and pumping (With the exception perhaps of Vlado Meller, who mastered St. Anger, Californication and Audioslave's first album )

Back to the topic:
An interesting thing I noticed on the Alanis Morissette DVD-A of "Under Rug Swept" is that the voice sounds way more compressed than the rest of the music.

Here is a sample: Alanis DVD-A Sample (23s, 2.3MB, flac)

with screenshot


The track is downmixed from the 5.1 version and has a trackgain of -4.25dB and all the instruments sound really good, but the voice is heavily compressed and just sounds somewhat thin. I just found it interesting that apparently the voice was recorded that way for the normal CD-version and the harm couldn't be undone later when a more dynamic take would have been needed.

This post has been edited by Vanishing: Jul 21 2005, 23:00

[tgoose]

So what's everyone's thoughts on the ethics of downsampling the better mastered DVD-A and making it available to people who own the CD?

[precisionist]

(With the exception perhaps of Vlado Meller, who mastered St. Anger, Californication and Audioslave's first album )

Confirmed. I've come across a lot of hopelessly squashed albums mastered by him.

So what's everyone's thoughts on the ethics of downsampling the better mastered DVD-A and making it available to people who own the CD?

I'm sure the complex ripping process is considered a "non-simple mean" of circumventing a copy protection and that's therefore illegal in the EU.

[ChangFest]

You're right, it was.  Boo.

No big loss, the 5.1 converted to stereo is great.

Man, I should buy the DVD-A and downmix to stereo. I'll throw away the CD then.

[Vanishing]

So what's everyone's thoughts on the ethics of downsampling the better mastered DVD-A and making it available to people who own the CD?

I'm sure the complex ripping process is considered a "non-simple mean" of circumventing a copy protection and that's therefore illegal in the EU.

Nevertheless it's a great way of showing your friends what non-squashed music sounds like.
I set up an ABX-session between a CD and a DVD-A version of a song so that I can easily switch between the two. Everyone I showed it to heard the differences at once and when asked preferred the unsquashed track.

[krabapple]

So what's everyone's thoughts on the ethics of downsampling the better mastered DVD-A and making it available to people who own the CD?

I'm sure the complex ripping process is considered a "non-simple mean" of circumventing a copy protection and that's therefore illegal in the EU.

Nevertheless it's a great way of showing your friends what non-squashed music sounds like.
I set up an ABX-session between a CD and a DVD-A version of a song so that I can easily switch between the two. Everyone I showed it to heard the differences at once and when asked preferred the unsquashed track.

er..wouldn't the *mixes* be different too? I'd be very surprised if a 6-channel remix , downmixed to 2-channel, sounded the same as the original 2-channel mix.

FWIW, from doing analog-to-digital captures of stereo mixes from various SACDs and DVD-As, sometimes the differences between the waveforms of the 'hi rez' and Redbook versions on the same disc are very small (e.g. the Rolling Stones SACDs) and sometimes they're rather stunning (Dark Side of the Moon SACD). Some hi-rez layers appear obviously jacked up, though not actually clipped (Yes' Fragile -- I didn't capture the DVD-V layer version for comparison, though).

This post has been edited by krabapple: Jul 28 2005, 21:10

[2Bdecided]

Nevertheless it's a great way of showing your friends what non-squashed music sounds like. 
I set up an ABX-session between a CD and a DVD-A version of a song so that I can easily switch between the two. Everyone I showed it to heard the differences at once and when asked preferred the unsquashed track.

Very true - if you compare squashed to unsquashed versions of something with the loudness matched it's very rare for people to prefer the squashed version.

That's a relatively modern phenomenon. Back in the day where compressors were used to match the loudness of unprofessional sounding varying loudness sources (e.g. someone talking, but moving around in front of the microphone so the level kept changing) the "squashed" version would sound much better.

It's a matter of degree. A little compression can really help in some circumstances. Compression used in the loudness war is almost always fatiguing and removes something from the listening experience.

Cheers,
David.

[krabapple]

Not a DVD-A, but an SACD, -- an example of probable mastering differences between CD and SACD layers. The example is 'Up the Hill Backwards" from the David Bowie SCARY MONSTERS SACD. All .wavs normalized to 0 dBFS

The CD layer was ripped with EAC; it was also digitized from the 2-channel analog outputs of a Pioneer DV-45a, as was the DSD layer. The soundcard was an M-audio 2496 and the digitalizing software was Audition (@ 32/88.2, then normalized, resampled and dithered to 16/44.1)

Up the Hill Backwards


Note that the A/D transfer and the rip from the CD layer look and measure substantially the same, indicating that the A/D transfer process was pretty faithful. Unless my player is doing something weird to CD vs SACD -- and it shouldn't as all delay, levels, BM etc were set to the equivalent of 'neutral' -- it looks like the CD layer is more compressed than the SACD layer.

This post has been edited by krabapple: Jul 30 2005, 09:21

[irishcrazy2005]

Okay, this may be a really stupid question, as I am new to a lot of this stuff. When you say look at the compression on the original CD, what kind of compression are you talking about here?

If DVD-A is simply recorded at higher sampling rates and larger bit-depths, what is being "compressed" in regular CD audio? Thanks for the help.

-Phil

[Frank Klemm]

I debated adding this to one of the existing DVD-A threads, but they're getting a little congested.

With DVD-A now rippable, let's use this thread to analyze DVD-A mastering and, if possible, how that mastering compares to the CD version of the album.  The primary difference will hopefully be that DVD-As are unaffected (or less affected) by the loudness race.

Here's a look at "Deadwing" by Porcupine Tree.

Is there a WAVstat analysis available of these two pieces of music?

See http://www.huennebeck-online.de/musik/stat/

Easiest way is to count the level distribution and send the result to me.

unsigned short Distr [65536] ;
signed short Samples [...] ;
long SampleCount = fread ( Samples, sizeof(*Samples), sizeof (Samples)/sizeof(*Samples), fpi ) ;
for ( i = 0; i < SampleCount; i++ )
Distr [ Samples [i] + 0x8000 ] ++ ;
fwrite ( Distr, 1, sizeof Distr, fpo ) ;

[Vanishing]

Okay, this may be a really stupid question, as I am new to a lot of this stuff.  When you say look at the compression on the original CD, what kind of compression are you talking about here?

If DVD-A is simply recorded at higher sampling rates and larger bit-depths, what is being "compressed" in regular CD audio?  Thanks for the help.

-Phil

"Comression" can be a misleading term, as it is used as the short form of "data compression" (mp3 etc.) but also "dynamic range compression", which is what we're talking about here. Dynamic range is the difference between the softest and loudest part of a song. In today's popmusic there's the problem that everyone wants to have the loudest recording, so the dynamic range ist compressed, with the result that the softer parts become louder and thus raising the perceived loudness.
This is a big problem because the lack of dynamics means also a lifeless and and flat sound. As a side-effect clipping and/or pumping can occur, which further deteriorates the sound. Apparently 5.1 mixes on DVD-As and SACDs are not affected by this "loudness race", so they're a welcomed alternative to today's CDs and a good chance to hear unsquashed music.

[CarlosTheTackle]

This is a shame isn't it. No doubt it won't be long before we see reviews touting the wonders of the extra bit-depth and sample rate, comparing how it sounds 'more lifelike' and 'more open' and 'exciting' than the CD equivalent, when in fact it's got nothing to do with the extra resolution - just better mastering practices.

Nifty ploy by the major labels there - if you can't really improve on the CD, just devalue it by making all mastering for it crap and make people think they need the new latest and greatest format.

Sad.