Back Forum Reply New

Released: DolbyProLogicII decoder that works! (proper channel separation)

With the new release of Aud-X (v. 1.2) we lounched a DPLII decoder. I spent tens of hours in Matlab to figure out how to spearate all the channels properly, so I hope you will  like the result .   
I was able to test the results on my, my wife and my friends and it seemed to be similarly effective, in terms of channel separation, as the original, hardware Dolby decoder.
All the tests were perfomed on the same HT amp.
Interestingly, they claimed that my algo sounds better in terms of linear distortions (the test noise played with the original DPLII decoder seemed to have distorted frequency characteristics...)
---------------------------------------------------------
But, ofcourse opinion all possible of users is very interesting for me, so I could use some suggestions to improve this algo.
I attach a test sound for ya (maybe it will be useful).
or download it from:
images/gu...dplii_dwmx.zip
=============================================
Just in case (probably for most this is more than obvious  ), here are some details how to test the algorithm
and compare it to other existing ones.

To perform a comparison with the use of external hardware home theatre system, please do the following:
1) Select in Aud-X DS filter config PCM stream compatibility.
2) Select in Aud-X DS filter config SPDiF output. The decoded (to 5.1) stream will be encoded to AC3 (448 kbps), and sent to the external amp. And you will be able to listen to the results of our algorithm
3) Select stereo output in Aud-X filter to send the original stereo (DPLII downmix) to the external amp.
4) Select in your amp DPLII decoder.
===========================
To perform a comparison with the use of other DS filters (e.g. FFDshow), please do the following:
1) To use other filters, please disable Aud-X (or disable PCM stream compatibility), and select an appropriate decoding option in the compared filter.
2) Enable PCM stream compatibility and use Aud-X 5.1 output
to decode the input DPLII downmix. Please make sure that the input stream is not already decoded to 6 channels by other filters, preceding Aud-X

=================

The channel order in the test file is
1) FL
2) FR
3) C
4) SL
5) SR

Hiho,

can plz chose another download source until you attachment willbe apporved -gt; rapidshare f.e.

THX


Originally Posted by daphyHiho,

can plz chose another download source until you attachment willbe apporved -gt; rapidshare f.e.

THX

oopz, unfortunately I would have to ask my colleague
who is responsible for our web site to put is somewhere...this
will take at least a week  

But I just tried, and I am able to download it...
If you still can't, I can send it to you via email.
(please provide your address)

~gt;see PM

Hi 3dsnar,

I wonder, could you provide a basic quot;speaker mappingquot; test sample file please?Cheers

temp. download  
(chose the free button)

I am sorry, I do not have access to this nice lady
saying in erotic way: 'fron left, front right, etc.'  
But seriously, please use the file provided by Daphy.
or
images/gu...dplii_dwmx.zip

It is a white noise and allows to see (hear) the linear distortion.
If you have other test files, it would be great if you could provide a link! Thank you.

Or PM me your email address and I will send it via email.

Sadly I'm unable to download anything from RapidShare these-days.... YouSendIt seems to work okay for me though.

Here's a link to my web site wher I have a number of 6Ch audio sources: -

51_Tes...est_Files.html


Originally Posted by SeeMoreDigitalSadly I'm unable to download anything from RapidShare these-days.... YouSendIt seems to work okay for me though.

Here's a link to my web site wher I have a number of 6Ch audio sources: -

51_Tes...est_Files.html

I have sent you an email.
But anyway, here is your 6 channel male speech (from your site) downmixed to DPLII.
Please test it.
images/gu...dplii_dwmx.rar


Originally Posted by 3dsnarInterestingly, they claimed that my algo sounds better in terms of linear distortions (the test noise played with the original DPLII decoder seemed to have distorted frequency characteristics...)

It also depends on the encoder amp; matrix used(emulations) unless you are using the Dolby H/W DPL II encoder which is intended for certified H/W decoders.
The complexity of the streams also affects the channel seperations significantly, so the mixed stream tests(6ch speaker test file + 6ch music) are necessary to be more accurate amp; practical.

The 6ch speaker test files(no concurrent channels) will produce almost identical good seperations unless the shifted phase image is broken(by conrainer codecs usually).

The assumed downmix matrix equations are:
Lt = L + 0.7071 C + 0.7071 LFE - 0.866 BL - 0.5 BR
Rt = R + 0.7071 C + 0.7071 LFE + 0.5 BL + 0.866 BR
with reference to this thread:
showthread.php?s=amp;threadid=57988
====================
The rest what you wrote is 100% true
(while this is DPLII)

However, the test files are good to start with.
If you have some nice complex test files, please
provide some links.
Thank you.

Well, I can just provide the links on how to encode DPL II mixes using Avisynth if you are not informed.
So again the very familiar(?) links :
showthread.ph...534#post788534
showpost.php?...amp;postcount=149

So we agree on that the standards to tune to are Dolby certified H/W decoders(amp; encoders) not the s/w encoder(amp; matrices, decoders) emulations...

I have no problem with creating downmixes
Simply, I am interested in users' opinion regarding various test files, including their own tests. I have tested it agains a certified hardware decoders.
---
Exactly! The target is to emulate (or possibly exceed a bit if possible) the certified hardware decoders, since they are relatively simple, while these days it is possible to use some complex DSP techniques.
So, I decided to do this, because
1) There are no free software DirectShow DPLII decoders which work.
2) There are plenty of movies with DPLII sound

But ofcourse, the DPLII will never be as good as the multichannel digital formats, such as AC3, AAC 5.1 or Aud-X,
thus I perceive this as a sort of limited substitute...

Pls don't get me wrong. I believe you can make the DPL II mixes no doubt!  
What I linked is an easy practical way if anybody is familiar with avisynth script and it's also because you did not specify the encoder you used.

But ofcourse, the DPLII will never be as good as the multichannel digital formats, such as ...
thus I do not perceive this as a competing solution.

Some recent results show that the AAC(he) actually provides decent 6ch discrete encoding down to 96khz VBR.
But if you want further compressions(ie. 40~96kbps range) with close(70~90%) spatial resonstruction, the DPL II is strongly considered a good candidate.

I still see the clear advantage of the DPL II for 6ch decoding, in addition to the lightweight portable use, but it's JUST MY OPINION!, with some background facts.

Ofcourse, I understand and agree with you.

DPLII is just a matter of application, needs, and equipment used
for decoging the stream, etc, etc.
Therefore I did decide to implement it.

Since our DS filter can work as a posprocessing unit
(when appropriate streams are chosen in the config tab)
you may create the DPLII stream with any format,
and then use any DS filter (such as FFDshow for example) decode it to 2 channel PCM,
which will be directed to our filter, which will do the DPLII decoding.

Initially (after the installation) the config of Aud-X decoder is set
to decode PCM and Aud-X compatible streams only. So in such
configuration this should serve as Aud-X decoder, or postprocessing unit for DPLII decoding, and for pseudo-surround
DSP processing (i.e. AC3 or standard MP3 will be handled by other filters).

I believe I made my point very clear about your aud-x dsfilter before.( you don't have to repeat it to me if you don't agree)

Because it does not support two-way compatibility( aud-x decoding-gt; ffdshow DSP+out ), I really really had a hard time to evaluate your efforts.
If you are unable to appreciate and live with existing others' efforts, probably I guess that's the reason why I can't see aud-x.

To be honest, I really want(ed) to appreciate your contribution...

In fact, what I wrote, reffers to the new release v1.2 (available from today), which has new options. But still, it can be used only
as the last filter (or non if you disable it) in the filter chain.
But I do not want to argue about this or force anyone to use it ofcourse.
I just wanted to explain, and help (on how to use the DPLII decoder embedded in the filter). That's all.

And the thread title is Released: DolbyProLogicII decoder that works! (proper channel separation) .. that's why I am in here.

OK. that is reasonable.

(to pass PCM via SPDIF) and have made the following observations: -

If I set MediaPlayer Classic to quot;Repeat Foreverquot; mode, when I play the quot;speechquot; WAV sample (you provided), it plays thru' the file quot;oncequot; and then stalls.

If I set Windows Media Player 10 to quot;Repeatquot; mode, when I play the same file, it plays thru' the file and repeats perfectly.

I did notice that the quot;iconquot; running in the system tray disappears quot;a good second or soquot; before the files end run-time. Would this cause any playback issues?Cheers

OK, we must investigate the media player classic issue.

The icon disappearing seems to be controlled by the MP10
(i.e. at the end of the stream it intializes the filter chain again, and again...).
I have experienced the same...This should rather not cause any playback problems.
¥
Back Forum Reply New