Software Dolby Pro-Logic II decoder?

[tebasuna51]

Here my test with new ATSurround for Foobar2000 0.9:
wav6 -> DPL II -> stereo_wav -> ATSurround Dec AS2 -> wav6'
Notes:
- wav6 is a Channel_Test (channels separated in time)
- wav6' obtained with Foobar Convert function
- DPL II with BeHappy and matrix M1 (see previous post) without LFE
- ATSurround Dec ASr have similar results
- Old ATSurround for Foobar2000 0.83 tested before with similar results
- Only amplitude are measured (without phase relation) then I don't put signs.

DPL II -> ATSurround Dec AS2
L R C BL BR
---- ---- ---- ---- ----
L' 0.67 - 0.06 0.30 0.03
R' - 0.67 0.06 0.03 0.30
C' 0.06 0.06 0.67 0.02 0.02
LFE' - - - - -
SL' 0.10 0.06 0.03 0.73 0.64
SR' 0.06 0.10 0.03 0.64 0.73
- Rear channels mixed
- Front-Rear channels mixed
- Very high rear global volume (compared with front volume)

And now, using Foobar Convert function also for stereo_wav:
wav6 -> ATSurround Enc -> stereo_wav -> ATSurround Dec AS2 -> wav6'
ATSurround Enc -> ATSurround Dec AS2
L R C LFE BL BR
---- ---- ---- ---- ---- ----
L' 0.67 - 0.06 0.15 0.05 0.05
R' - 0.67 0.06 0.15 0.05 0.05
C' 0.06 0.06 0.67 1.71 - -
LFE' - - - - - -
SL' 0.10 0.06 0.03 0.07 0.68 0.72
SR' 0.06 0.10 0.03 0.07 0.68 0.72
- Rear channels mixed
- Very high rear global volume (compared with front volume)
- Very high LFE volume (and can't disable LFE input in ATSurround Enc)

@3dsnar
Check if ATSurround is active at File -> Preferences -> DSP Manager in Foobar.

[Rockaria]

wav6 -> DPL II -> stereo_wav -> ATSurround Dec AS2 -> wav6'I read a user say AS2 produces too much mixes/surrounds in their forum.
http://www.andrewlabs.com/forums/forum-1.html

[3dsnar]

I've tried only the winamp plugin.
I will test the fb2k plugin then.
But from Tebasuna's experiments
I see that it does not work as it should...

[radosuaf]

Read the topic - I don't think it's of any help for me... Card is 5.1 and so are the speakers. One thing is the center speaker and the other is the stuttering.

[3dsnar]

Do you have this problem with playing DPLII files (i.e. AC3, MP3 or PCM)?

[radosuaf]

Tried MP3s with Windows Media Player 10 yesterday and it worked really nice. Don't know what's the problem...

[3dsnar]

Hmm, unfortunately I do not have your hardware to test it :(
Maybe it consumes to much of your CPU (and this is the reason for stuttering?)
Did you manage to use other filters succesfully (e.g. FFDshow)?

[specise_8472]

Why can't we just live with the fact that DPLII requires 90 degree phase shift?
This is the spec, and anything else is just out of spec.

BTW DTS uses 45 degree phase shift in the rears when matrixing in the Center Surround channel.
QUOTE:
Phase shift ES channels – Works only with ES source files, where it allows for phase shifting of Left and Right Surround by ±45° prior to addition of Center Surround channel.
Note: DTS strongly recommends Phase shift be set ON during 6.1 ES encoding for film. For music, you may wish to experiment with both settings.
END QUOTE:

[scharfis_brain]

Why can't we just live with the fact that DPLII requires 90 degree phase shift?
This is the spec, and anything else is just out of spec.

It may be the spec for DPL1 encoding. But I've never seen official specs describing DPL2 downmixing properly.


If you have sources that do so, I'd be interested very much in them.

[specise_8472]

It may be the spec for DPL1 encoding. But I've never seen official specs describing DPL2 downmixing properly.


If you have sources that do so, I'd be interested very much in them.

Somwhere on my cluttered gigs of HDD I have the Patent for DPLII (Not specifically mentioned as such - but reading leaves only that impression from the technology used). I will dig it out and let you know.

[3dsnar]

The thing is that it might be recommended, but without it the certified decoders still work. The questions is if the 90 deg. phase shifts during encoding (especially for some modern 5.1 audio) bring anything extra in terms of decoding. This requires (even informal) listening tests, and would be interesting to find it out.
I'll try to provide some audio examples (one with 90 deg shift and one without).

[3dsnar]

OK, I have prepared the samples:
Please follow this thread (poll) and decide which ones
sound better:
http://forum.doom9.org/showthread.php?t=112122
Cheers,
3d

[Rockaria]

Lt = FL{0°} + 0.7071 C{0°} + 0.7071 LFE{0°} + 0.866 BL{180°} + 0.5 BR{180°}
Rt = FR{0°} + 0.7071 C{0°} + 0.7071 LFE{0°} + 0.5 BL{0°} + 0.866 BR{0°}
and
Lt = FL{0°} + 0.7071 C{0°} + 0.7071 LFE{0°} + 0.866 SL{90°} + 0.5 SR{-90°}
Rt = FR{0°} + 0.7071 C{0°} + 0.7071 LFE{0°} + 0.5 SL{-90°} + 0.866 SR{+90°}

You might want to add two more models to fully reflect the discussions in the related threads:
Lt = FL{0°} + 0.7071 C{0°} + 0.7071 LFE{0°} + 0.866 BL{-90°} + 0.5 BR{-90°}
Rt = FR{0°} + 0.7071 C{0°} + 0.7071 LFE{0°} + 0.5 BL{90°} + 0.866 BR{90°}
and
Lt = FL{0°} + 0.7071 C{0°} + 0.7071 LFE{0°} + 0.866 SL{180°} + 0.5 SR{0°}
Rt = FR{0°} + 0.7071 C{0°} + 0.7071 LFE{0°} + 0.5 SL{0°} + 0.866 SR{180°}


Please test them with your dolby certified decoders
and decide which ones sounds better (eg. A vs B version).
The poll question is related to fifthelem_A.mp3 and fifthelem_B.mp3,
while speech_A.mp3 and speech_B.mp3 should sound identical.

You might also want to test with original 6ch(mp4 format) vs dpl II mix(music + speaker test clip), to compare reasonably.
The FFDShow can switch between PCM/AC3 in digital out mode on the fly making it easy to compare.
As I said many times, the speaker test clip is most generous on any models(more than 95% of the seperation quality).

Also the matrices values(Ls1,Ls2,Rs1,Rs2....) might affect the seperation quality by the phase shift degree change.
Comparing the Wikipedia matrix with the current one would be reasonable( also making it reasonable using the variable reference than the values).

So far, the fifthelem_B showed noticeable better seperation, but the test is not setup to compare the seperation fidelity(original vs DPL II).

[3dsnar]

I think to other models would give similar quality,
to the 90 deg. shift and to the 180 deg. shift respectively.

And it would complicate the comparison, so I decided to stick to the two selected downmixing methods.
The difference between wikipedia and the provided weights, should not affect the results.
(the difference is more related to levels in the decoded channels, but it is very small)

Please make your vote in the poll.

I also provided the input (reference) file.
http://forum.doom9.org/showthread.php?t=112122

Cheers, 3d

[Rockaria]

I think to other models would give similar quality,
to the 90 deg. shift and to the 180 deg. shift respectively.

And it would complicate the comparison, so I decided to stick to the two selected downmixing methods.

Please make your vote in the poll.

I also provided the input (reference) file.
http://forum.doom9.org/showthread.php?t=112122

Cheers, 3d
No, it's just your thoughts. If 'no listenings(reading)' is the trend of modern conversation, I think I've done my duty.

[3dsnar]

Rockaria, if you find valuable testing all reference
downmixing methods, please provide the necessary test samples,
and organize a new poll

My thougths = my intuition = knowledge and experiance.
Since depending only on certainty is very energy/time consuming,
I have to rely partly on my intuition.
To be more specific, it would take extra effort to prepare more test samples :)
Cheers, 3d

[Rockaria]

My thougths = my intuition = knowledge and experiance.
So the second model(reference 3) purely came out of your thought & career!:scared:
I believe what you did is just(although I appreciate the latter):
. asking the matrix format to your fomula style
. implemented a s/w logic of 90 degree phase shift.

from :

@Rocaria.
Well, in fact I have no idea how the original DPLII decoder works...
I knew how to downmix (thanx to your post, by the way ),
and I have some DSP knowledge and experience, so figured out
how to resolve that. Since I was thinking in terms of the latest
DSP techniques, possibly my approach is a bit more modern that Dolby's approach... And this may result in better quality

I believe you might also have to exclude 'DPL II' from your current dsfilter because it's apparantly misleading. Isn't is just an active matrix decoder, modern or not?

If you don't like to perform the relatively reasonable comparisons and drive the topics for your desires, do it in your own thread. I won't touch it.

[3dsnar]

Hmm, you are very emotional.
Take it easy, please :)
--
The (my) idea behind the comparison is to see if the 90 deg.
phase shift brings anything extra (or maybe it is opposite)
to the already tested (with results reported) downmixing
equation:
http://forum.doom9.org/showthread.php?s=&threadid=57988

So it is not my equation.

Will see.
--
I think it is reasonable to call it DPLII decoder (in our DSfilter),
because it attempts to upmix DPLII downmixes.
And IMHO this is the simplest way to describe its functionality.
(by calling it DPLII decoder. Most people do not know, nor care how it works. They just want to get multichannel signal from the downmix. That is all.)

It would be nice of you to vote.
If you would decide to compare all 4 ref. downmixes,
I would also vote.

Cheers, 3d.

[Rockaria]

No, I am absolutely cool.:cool: you are driving.
Your linked thread(4~5 times so far) unfortunately does not contain my matrix(reference 3== your second model) or similar formula, if you want to regard it as your dictionary.

From absolutely 'no idea' on DPL II in the beginning of this thread, you've made a remarkable progress in figuring out the DPL II model, all of a sudden, out of no where, by yourself! It's not a conversation or cooperation, not compliant again.

What I count with the Atsurround plugin(& FFDShow...) is, they does not claim it DPL II compatible, at least.

/no deadhorse pls.:)

[3dsnar]

No, I am absolutely cool.:cool: you are driving.
Your linked thread(4~5 times so far) unfortunately does not contain my matrix(reference 3== your second model) or similar formula, if you want to regard it as your dictionary.

From absolutely 'no idea' on DPL II in the beginning of this thread, you've made a remarkable progress in figuring out the DPL II model, all of a sudden, out of no where, by yourself!
It's not a conversation or cooperation, not compliant again.

What I count with the Atsurround plugin(& FFDShow...) is, they does not claim it DPL II compatible, at least.

/no deadhorse pls.:)
Cool as ice... :rolleyes:

The second equation comes from your post:
http://forum.doom9.org/showthread.php?t=111603&page=2
I only ajusted some of the weights to make it complient with the first one.

I have never seen any Dolby's specs on how to decode the DPLII matrix, but on the other hand I am pretty much sure that they used the same general concept which I did (I do not see any other way to decode the stream).
Therefore my speculations and assumptions may be correct.

That is all I wanted to say.
Peace, 3d

[Rockaria]

Maybe I am more inclined to the male-characteristics when engaged in the logic?:rolleyes:

The DPL II matrix on the Wikipedia looked to me more formal than the matrices in other threads, although I settled down to the same valuse as I described in the original thread for s/w 180 degree shift encoding.
That's why I said we need to test with the Wikipedia matrix for 90 degree shift, if you actually respectfully read my posts at all.
http://en.wikipedia.org/wiki/Dolby_Pro_Logic

http://forum.doom9.org/showpost.php?p=782907&postcount=40
You adjusted the values here? I already used the same values there!

The linked thread also contains the test results and analysis on the reference 2,4 models and varying results by the types of contents( speaker test file, 5~6ch music sources and mixed clips), which is the reason why I said we need to test the seperation fidelity as I mentioned above.

In this thread, I explained why the 180 degree shfits(based on various threads including your link) can never be as good as original 90 degree shifts(by Dolby) to deal with the the iiresponsibly(with no basis) repeating assumptions.

Pls figure out what has been actually misleading us to get the correct implementation of the s/w DPL II emulations earlier.
The seemingly convincing test results with the most generous speaker test clip, no seperation fidelity test, no theoretical background & reasoning, irresponsible parroting with no idea..only to refuse or attack., i guess, are included.

The reference 1 might also show the reasonable decent results, if you haven't tested it yet.
But if you have and excluded it because of no good results, Why don't you just say that?

So if you want to build a new house, start it from the scratch or list what resources/references you are using
, which is the widely accepted way to enjoy the conversations with no loss, in full respects.

[radosuaf]

Hmm, unfortunately I do not have your hardware to test it :(
Maybe it consumes to much of your CPU (and this is the reason for stuttering?)
Did you manage to use other filters succesfully (e.g. FFDshow)?

Hmmm... WinFast PVR takes about 14% of CPU power, CPU is Barton Mobile @2222 MHz so I guess it should be sufficient for such purposes.

[3dsnar]

Hmm, yes...
Please let me know if you succeeded in connecting
other DS filters (e.g. FFDshow).
:thanks:
3d

[ramabadr]

Hello,

Is there a standard published version of the DPL II decoding (up-mixing) algorithm? I have read the article "Pro Logic II matrix decoder melds multimedia" by Roger Dressler of Dolby Labs. It gives the general concepts but I would like more specifics to be able to implement the algorithm in software. Members of this forum seem quite knowledgeable on this topic - can some one help? Thanks.

- Ram