Here is what I think is a fully digital copy from my Santana DVD-A for Tigre or someone else to test. I don't really know how you would tell without having a ripped aob file to compare with. But in any case here is a stereo clip at 24/96. If it is a true digital copy I will be very happy then I can take my DVD-A music on the road with me


Edit: .wav attatchment replaced by wavpack version - saves lots of space/bandwith - tigre

Hi sayersc, welcome to Hydrogenaudio!

Thanks a lot for your effort. I loaded your sample (beginning of Santana - Supernatural (05) Smooth) - with CoolEditPro.

1st I checked the 'silence' in the beginning of the track - there's a short part with 24bit LSB dither/noise = +/- 1 amplitude.

But on the other hand there's no content > 20kHz in spectral view / frequency analysis besides a few clicks that are obviously (ripping?) errors.

I own the CD version, unfortunately I can't find the CD right now, but the Musepack insane version is on my HDD, so I compared. Here's the frequency responses of DVD-A vs. CD version (same passage, volume adjusted, DVD-A version resampled to 44.1kHz, only left channel):

!!! The red graph is CD->MPC version, the green one is DVD-A !!!

So obviously the 'DVD-A' version has a 8-16kHz lowpass!
Besides this, I noticed this: The CD version is recorded too loud as many recent pop CDs (replaygain values -7.5/-7.8dB), on basedrum hits there's heavy clipping (nearly horizontal line with ~ full scale sample values > 100 samples in a row). The 'DVD-A' version is amplified by ~ -5dB compared to the CD version but has almost the same horizontal lines ("almost", because the lines are somewhat smoothed, probably because of lowpass).

Examples:


CD->MPC version


'DVD-A' version

Conclusions:

• The 'DVD-A' version seems to be 24bit resolution because of the LSB noise. This noise might have occured during 'ripping', though (IIRC there were some reports that some newer Windows versions do some LSB changes on directsound/waveout playback - this could also explain the LSB noise during silence).
• No audible content > 20kHz and the lowpass are strange, so either
  • this 'DVD-A' contains an upsampled CD 'rip' with lowered volume that has been lowpassed to make it sound more vinyl-like or
  • your DVD-A contains the same songs as DVD-V version (= video with AC3 or DTS audio stream) for compatibility reasons and you've ripped that one.

As the lowpass is easily audible you could try to check this (if you have a standalone DVD-A player) - compare playback through WinDVD + souncard with standalone DVD-A player playback (use same headphones /amp+speakers).

Cheers tigre

Another thing Moneo just told me on IRC: When checking the file with a Hex Editor you'll see that there are xx xx 00 xx xx 00 patterns which means, the '24 bits' are 16 bits padded with zeros to 24.

The question remaining is: What is the source for this "non-audiophile'ish" behaviour here - the DVD-A or the Software? Maybe you could create another file from another more 'audiophile' DVD-A?

I am noticing that DVD-A discs have so many different variables in the way they are recorded that it is difficult to pin down the variables.

I will try Queen which has 24/96 PCM stereo in the Video Folder! (Is that even allowed?)
24/96 DTS in Video folder and 24/96 6 channel in the Audio folder.

I am nearly 100% positive that the Santana recording was from the 24/96 DVD-A as the info in WinDVD says that is the stream that is playing and it obviously sounds different than when I play from the video folder(audible difference).

Funny that drop off starting at 8000 Hz, I don't seem to be able to hear the difference between the wav file and the WinDVD out to my headphones. Perhaps my headphones are not sensitive enough.

Perhaps Wavelab will record at 24 bits and add the padding even though the source is 16 bits.

If you can give me some suggestions to try I can try direct sound out of WinDVD or Waveout or Auto. The Santana clip was done at Directsound to Prodigy and sent to Wavelab. Also tried demo of N-track but the sound is definately different on the recorded wave compared to listening with WinDVD directly.

I am thinking maybe WinDVD only outputs 16 bit? without the Audigy

Yes. 24/96/stereo LPCM is allowed and such DVD-Vs exist, e.g. by Chesky. To convert to .wav "LPCM 24bit processor" from here is necessary.

24/96 DTS in Video folder and 24/96 6 channel in the Audio folder.


I attatched a CD->mpc version of the 1st 20 seconds of the track to this post - compare if you want ...


AFAIK Adobe Audition trial version is fully functional - and capable of recording higher resolutions than 16/48. Instead of routing the signal through the soundcard, you could also try NTONYX virtual audio cable.

Thanks for the hints tigre but I think WinDVD simply does not put out more than 16 bits unless it finds the Audigy. Both Wavelab and N-Track will do Multitrack 32/192 so it is not those programs limiting the recording. I downloaded Hexedit and did many more tests and the all my samples have the padding zeros. Sigh.

I am aware of the Virtual Audio Cable as I am the one that suggesested it in Aftredawn forums. VAC will only pass 16 bit output from WinDVD although it is capable of 24 bit.

My Prodigy 7.1 will pass whatever signal it receives on to the reording/effects program using ASIO in digital form without going through the DACs . So in effect it is performing the same function as Virtual Audio cable. Also the same as doing a digital loopback with two soundcards.

It appears that WinDVD only passes 16/96 max bit unless it finds Audigy2. I used regspy to see what is happening in registry when WinDVD starts in DVD-A mode. Very interesting and very complex.

At this point I am going to give up on getting more than 16 bit out of WinDVD until a skilled person can find a switch to turn it on for soundcards other than Audigy2. The Prodigy 7.1 is capable of 24/192 on eight channels and is more than capable to the task.

Even if there is a way to get full resolution out of WinDVD, there will be errors in recording the stream, as you have already pointed out.

I don't know if there is such a thing as error correction on PC DVD drives for DVD-A yet?

Most DVD-Audio discs have an AC3 version of the recording so the disc is backward compatible with DVD-Video players. Because the LPCM is encrypted, and possibly even losslessly compressed with MLP, I would expect that what was ripped was an AC3 version, not the 24/96 LPCM.

DigitalMan, I am fully aware of the Video vs audio folder stuff. A good example to test wether WinDVD (with DVD-A capability) is playing the Video vs. Audio folder is try to record the stream from Queen "The Game" This disc has DTS(5Channel 24/96 Surround),
24/96 LPCM Stereo in the video Folder and DVD-A 24/96 6 channel in the audio folder.

If you play the stereo track you will get only 2 channels output(check whatever method works).

This disc is a good check for WinDVD because there is only one possible version of the Stereo track.

Are you suggesting that WinDVD might pick the DTS track and output only two channels if you choose the Stereo option when WinDVD starts, If you don't have the required Audigy 2 Card?

In any case all three options on this disc ar 24/96 and I can only record 16 bits.

I am almost convinced WinDVD only puts out 16 bit unless Audigy2 card is installed due to obvious digital copying possibilities. It still sucks as I have no way to play DVD-As in the vehicle and I can't make AACs of the higher bit depth available from DVD-A unless I do it analogue.