difference between clean stereo and dolby prologic/prologicII

[aprunic]

Hello all.

I've been testing besweet's capabilities of creating dolby prologicII compatible wav's, and when I played them on my surround system using av receiver, the sound is quite good.
Not as good and distinct as with true ac3/dts 5.1 encodings, but quite nice...

But, I've come across a question I can't answer my self: how is the prologic/prologicII wav file different from the standard stereo wav, considering left and right channels?
Or, how are original stereo left&right channels different from left&right front channels in prologic/prologicII encoded stereo file?

Why do I ask? Well, for one reason - if I play regular stereo file on any stereo equipment, I will have the same left&right channels.
But, if I play PL/PLII encoded file on an equipment that does not know how to use extra info (a regular stereo cd+amp), what will I hear on the only 2 speakers I have?
How much and how bad will the sound be changed? Will it matter at all?

Greetings, Andrea

[ursamtl]

Audio that's correctly encoded for Pro Logic I & II decoding is designed to be "downwards compatible" with stereo equipment. In others words, when you play it back on a regular stereo system, the extra sonic information is encoded in such a way that you don't hear it. If you check the Dolby web site and read some of the tech papers available there in PDF form, the whole process is explained.

Happy experimenting. Be careful, it can become addictive! :)
Steve.

[BigDid]

Hello Andrea,

There has been a thread I contributed to related to your questions here:
http://forum.doom9.org/showthread.php?postid=595940#post595940
even if it is about mp3, it can apply to dpl, dpl2 in general.
As suggested, you can also use the search fonction intensively to find more infos on the subject.

Good reading :)

Did

[aprunic]

Thanks guys for your answers, sorry I didn't reply sooner, I was rather busy and ill for a while...

So, what you're saying is that I could recode all my pure stereo CDs into DPLII, and if i take them and play them on some of my friend's regular stereo system, it will sound just like the original stereo CD?
No loss of information, no phase shifts, no loss of volume or coloration or something else on the front speakers, no getting purple dots on my back and a green tail on my neighbour's dog, right?

If that's so, then it is wonderful! That gives a lot of work ahead, just like you say, ursamti.. ;)

OK, now I'll go and dig out some docs from Dolby's site and try to see what I can understand from them...

Thanks.

[daphy]

general question:

are there any ways to encode to Dolby ProLogic IIx?

[aprunic]

Originally posted by BigDid
Hello Andrea,

There has been a thread I contributed to related to your questions here:
http://forum.doom9.org/showthread.php?postid=595940#post595940
even if it is about mp3, it can apply to dpl, dpl2 in general.
As suggested, you can also use the search fonction intensively to find more infos on the subject.

Good reading :)

Did

I've read through this thread, and come to a conclusion that PLI/PLII encoded stereo file can not be the same as the stero file, because it has more information it it.
A quick formula excerpt for PLI matrix:

Lt = 1.0*L + 0.707*C - 0.707*Ls - 0.707*Rs left-total
Rt = 1.0*R + 0.707*C + 0.707*Ls + 0.707*Rs right-total

As we can see, final stereo file is actually a combination of a couple of signals, and even though it keeps all the information from all surround channels in it, it can not sound the same as a pure stereo, non PLI encoded file, because pure stereo file is recorded with two channels while PLI is encoded with 5 channels as input.
And there is no way one can determine the correlation between the sound levels of information that is present in surround channels, when they are recorded in surround mode, and the same information when they are recorded in pure stereo mode.

Or, in other words, positioning, phase and volume of some instrument that is placed in an orchestra somewhere in the left middle is not going to be recorded the same way if you use just 2 channels for recording and if you use 6 channels for recording (final output, that is).

Now, am I talking jibberish, and being stupid, or not?

And is this difference in the sound something that really matters, can it be heard and how?

If final Lt and Rt are, on the other hand, having all the information in the right way as the L and R channels in stereo (or at least very similar) then PLI/PLII encoded file can be played on a regular stereo system without any sonic disturbances, and then we should reencode all of our stereo cd's! ;)

Anybody having any more good thoughts on the subject?
We're just warming up, folks! :)

Cheers.

[ursamtl]

I think the problem is that you're talking about two different things here. The formula for the DPLI matrix assumes that there is distinct center and surround information. If you just take your regular stereo CDs and recode them to DPLI or II, you'll get exactly the same thing because all you'll have is the original L and R signals. If you somehow generate C, Ls, and Rs channels, then you'll have something to encode and it will sound different from the original stereo signals.

[aprunic]

Originally posted by ursamtl
I think the problem is that you're talking about two different things here. The formula for the DPLI matrix assumes that there is distinct center and surround information. If you just take your regular stereo CDs and recode them to DPLI or II, you'll get exactly the same thing because all you'll have is the original L and R signals. If you somehow generate C, Ls, and Rs channels, then you'll have something to encode and it will sound different from the original stereo signals.

Noooo, DPL decoder does not care, as other threads said, what's the stereo stream it processes, and it does not differentiate between the regular stereo and DPL/DPLII in the terms of input signal.
The decoder uses whatever information it gets from the player, and then does it's job the best it can.
If the input stream is pure stereo, there is not much of needed phase shifted and mixed signals and the output channels are not distinguished very much.
But, if the source was encoded in the right way, then the decoder's input stream has enough information for it to recreate C, LFE and S (Sr,Sl) channels and bring back original L and R channels to the front speakers.

That's at least how I understood the process...

Now, if I take a regular CD and recode it into a 5.1 stream (which means I get 6 separate and distinct channels) and then use those 6 channels to create DPL/DPLII 2-channel wav, I am not using just original L and R to create DPL/DPLII output which has Lt and Rt, but a real 6 channels that are derived from original L and R channels through a process of summing, phase changeing, frequency filtering etc, much just like a real hardware 5.1 encoder would do, if it was given all 6 inputs from 6 microphones, for example...

So I am having all 6 channels to encode with, but I need to know how do encoded Lt and Rt sound on a plain stereo system, which has no DPL/DPLII decoder and can not recreate L, R, C, Sl, Sr.
How different are Lt and Rt compared to original L and R?

That's the question and the problem here...

[ursamtl]

Ok then, that's diffeent. Your original question, as I understood it, was if you "encode" regular stereo material as DPL I or II and then play it back on a regular stereo, will it sound the same as if it was played back directly. Using the encoding matrix you posted, the result should in theory be the same because there is no C or S signal if you encode just the L and R.

However, if you take regular stereo material and play it back through a DPL I or II decoder, you will in fact hear sound in your center and surround speakers. the C will be the sum of the two channels, and then surround will be derived from the difference between the two channels. Depending on your source material, the results can be quite good.

However, my answer to your original questions stands. Encoding 2ch stereo using a Dolby matrix and playing it back on a regular 2ch stereo should sound the same. If you derive or generate center and surround info or else artificially create it using reverb or similar effects, and then encode the material, obviously it will sound different when you play it back on a regular stereo system.

[aprunic]

Originally posted by ursamtl
However, if you take regular stereo material and play it back through a DPL I or II decoder, you will in fact hear sound in your center and surround speakers. the C will be the sum of the two channels, and then surround will be derived from the difference between the two channels. Depending on your source material, the results can be quite good.

Yup, I play quite a lot of my regular audio cd's this way, using DPLII Music decoding on my av receiver, and it does sound quite good on a lot of materials, especially on electronic with wide sperataion and panning effects, such that J. M. Jarre has.
On the other hand, MP3 files when played this way, suck most of the time, because, even though I create them always using full stereo, and not M/S encding, I lose channel separation and panning quite a lot, and DPL decoder can't do much about it during reproduction...


However, my answer to your original questions stands. Encoding 2ch stereo using a Dolby matrix and playing it back on a regular 2ch stereo should sound the same. If you derive or generate center and surround info or else artificially create it using reverb or similar effects, and then encode the material, obviously it will sound different when you play it back on a regular stereo system.

Now that was the question and the answer!
I did ask about that - how will DPLI/DPLII encoded material, made from 6 channels, sound on a regular sound system with just 2 channels.

So, what do you think will be the difference?
I tried it, and it sounded OK, but I didn't devote much time to it, and I do not have some equipment I could use to make some measurements (don't event have the idea what to measure... :)).
That's why I'd like someone here who has knowledge of the principles and the mathematics/electronics in encoding/decoding of DPL materials to try to explain how "bad" DPLI/DPLII material will be when played on a regular stereo system.

Cheers.

[ursamtl]

Originally posted by aprunic
Yup, I play quite a lot of my regular audio cd's this way, using DPLII Music decoding on my av receiver, and it does sound quite good on a lot of materials, especially on electronic with wide sperataion and panning effects, such that J. M. Jarre has.
On the other hand, MP3 files when played this way, suck most of the time, because, even though I create them always using full stereo, and not M/S encding, I lose channel separation and panning quite a lot, and DPL decoder can't do much about it during reproduction...
Actually M/S encoding might produce better results for DPL decoding. You see the Center channel is basically taken from the M and the surround from the S. The left and right channels are then derived by M+S and M-S. Mind you, this gives a mono surround. DPLI gives you just that. DPLII uses a steering logic and some phase work to give the illusion of separation in the surrounds by routing some of the front channel material to the back. So give M/S MP3s a try and what you get.


Now that was the question and the answer!
I did ask about that - how will DPLI/DPLII encoded material, made from 6 channels, sound on a regular sound system with just 2 channels.

So, what do you think will be the difference?
I tried it, and it sounded OK, but I didn't devote much time to it, and I do not have some equipment I could use to make some measurements (don't event have the idea what to measure... :)).
That's why I'd like someone here who has knowledge of the principles and the mathematics/electronics in encoding/decoding of DPL materials to try to explain how "bad" DPLI/DPLII material will be when played on a regular stereo system.

Cheers.
If the matrix you posted is used, the material should sound similar but with somewhat decreased stereo separation. This is because the Center channel is mixed in with the regular stereo. The surround material is phase shifted and then one channel is inverted to cancel the other out. Think of it this way: in the matrix you posted,Lt = 1.0*L + 0.707*C - 0.707*Ls - 0.707*Rs left-total
Rt = 1.0*R + 0.707*C + 0.707*Ls + 0.707*Rs right-totalthe - 0.707*Ls cancels out the + 0.707*Ls and the - 0.707*Rs cancels out the + 0.707*Rs. However, the 0.707*C is added to each channel, effectively increasing the M or mid content in each channel by 3dB.

[aprunic]

Originally posted by ursamtl
Actually M/S encoding might produce better results for DPL decoding. You see the Center channel is basically taken from the M and the surround from the S. The left and right channels are then derived by M+S and M-S. Mind you, this gives a mono surround. DPLI gives you just that. DPLII uses a steering logic and some phase work to give the illusion of separation in the surrounds by routing some of the front channel material to the back. So give M/S MP3s a try and what you get.

OK, I'll make one nice mp3 that has worked fine in the upmixing I've done so far, and make it in two versions, pure stereo and m/s version, and then try to play it as a mp3 but with PLII decoding.
Then we'll try to "feel the difference"..... :cool:



If the matrix you posted is used, the material should sound similar but with somewhat decreased stereo separation. This is because the Center channel is mixed in with the regular stereo. The surround material is phase shifted and then one channel is inverted to cancel the other out. Think of it this way: in the matrix you posted,Lt = 1.0*L + 0.707*C - 0.707*Ls - 0.707*Rs left-total
Rt = 1.0*R + 0.707*C + 0.707*Ls + 0.707*Rs right-totalthe - 0.707*Ls cancels out the + 0.707*Ls and the - 0.707*Rs cancels out the + 0.707*Rs. However, the 0.707*C is added to each channel, effectively increasing the M or mid content in each channel by 3dB.

Well, don't take me by the word (letter) because I copied it from another posting somewhere on these forums, and this formulae might be just a simplification of things, so I used it just as an explanation aid.
None the less, even if it is accurate (at least for DPLI), then you're saying that on a regular stereo system, DPL encoded file will actually be more "in the center" than the original because of an added center channel made from original L and R channels, and therefore the image will be narrower, right?

So, if the original stereo recording is concentrated in the middle (unfortunatelly, too many of them are like that), this way it WILL get worse because center components of the sound ambience will prevail over L and R sounds.
On the other hand, if the original recording is done nicely, with a good stereo separation, after encoding it in DPL and playing it like that on a stereo system, center components will just get amplified twofold, but since original L and R are good, the overall sound will not be that bad, and may even become better because of the appearance of a pseudo center channel which may even the sound balance from L to R speaker, thus giving it even nicer and richer sound...

This is what I conclude from your words, and also that a lot will depend on the material used.
For example, electronic music with a lot of effects and L-R panning will do just fine (and be encoded just fine as well) while vocal music, which is mostly in the center of a recording, will be bad either way.

So, I won't be upmixing my Suzanne Vega CD, while all of my J.M.Jarre will get a dts surround version as soon as I can find the time to do it.... :)

Anybody having additional thoughts?

Cheers.

[ursamtl]

Yes, you got the jist of it. As you said, it depends on the material. I've been playing around with this stuff for awhile now and I've found that material has a huge impact. That's why I added a width correction control when I wrote the V.I VST surround plugin. Sometimes the original mix is too narrow to derive a decent surround effect. Other times, it's too wide to provide solid soundstage imaging.

You might want to give the V.I approach a try as well. Combine it with a good impulse response loaded into the SIR plugin for the rear channel and you'll have an amazing sense of space!

[aprunic]

Originally posted by ursamtl
Yes, you got the jist of it. As you said, it depends on the material. I've been playing around with this stuff for awhile now and I've found that material has a huge impact. That's why I added a width correction control when I wrote the V.I VST surround plugin. Sometimes the original mix is too narrow to derive a decent surround effect. Other times, it's too wide to provide solid soundstage imaging.

You might want to give the V.I approach a try as well. Combine it with a good impulse response loaded into the SIR plugin for the rear channel and you'll have an amazing sense of space! And in case the Eye of Sauron is watching, I don't make any money from V.I so I'm not promoting it for personal gain. ;)

Aaaaahh yeeesss, I've seen somewhere these two plugins, but forgot where and what are they for.
So, why do I need them, what can they do, where do they "plug-in" and where to find them?

Oh yes, is anybody willing to share his/hers methodology of producing surround material? What do you choose for LFE cutting freq, what filtering, what are the level differences between the channels etc etc?

Might be a nice add-on to this little discussion of ours... :)

Greetings to all.

[ursamtl]

I think you'll find everything you need to know in
GUIDE: V.I Stereo to 5.1 Converter and II Surround Generator VST plugins by UrsaMtl (http://forum.doom9.org/showthread.php?s=&threadid=85446).

On the forum we've mainly used it with Plogue Bidule, an amazing program I'd highly recommend to anyone. However, you can use V.I with Cubase SX or Nuendo. You can also use II in most VST hosts. II only gives you the surround processing from V.I but deriving the front center and LFE can be done a variety of ways. V.I's LFE is cut off at 60Hz and is designed just to boost these extremely low frequencies when they're lacking, generally on older material.

Steve.

[You Know]

in conclusion if I have a Dolby Stereo Surround ac3 extracted from dvd, converted to wav with .... maven3d pro or besweet, edited with ... audition (for example synced with Japanese track), resaved into Stereo ac3 again, original surround are preserved or not?

[ursamtl]

If the channels are discrete Dolby Digital AC3 and all steps are carried out carefully, yes. By the way, "Dolby Surround" is normally used to refer to the original matrixed method Dolby marketed to movie theatres way back when. It's basically a simple design based on a Hafler circuit from the 70s.

[You Know]

on the web site material are marked as "Dobly Digital Surround 2.0"

i have obteined a stereo channel wave file.

which tools and setting i should use to get again a ac3 Dobly Digital Surround 2.0 from edited wav(the only modify are only cut/silence, no other effect applicated)?


EDIT: after read a few papers on dolby site, when i convert ac3->wav i should select Lt(left total)/Rt(right total) channel routing and not L/R channel. After that with a tool, processing this Lt/Rt to obtain four initial channel L,R,C,S. Is correct? Which tools permit to do this?

[ursamtl]

Originally posted by You Know
on the web site material are marked as "Dobly Digital Surround 2.0"

Yed, that's right. "Dolby Stereo Surround," as you called it last message is different from "Dolby Digital Surround 2.0."

i have obteined a stereo channel wave file.

which tools and setting i should use to get again a ac3 Dobly Digital Surround 2.0 from edited wav(the only modify are only cut/silence, no other effect applicated)?
If you have one of the commercial AC3 encoders (Soft Encode, Sony, Surcode), then use it. Otherwise, there is a free AC3 encoder that comes with the latest Beta of HeadAC3he. See http://needfulthings.webhop.org.


EDIT: after read a few papers on dolby site, when i convert ac3->wav i should select Lt(left total)/Rt(right total) channel routing and not L/R channel. After that with a tool, processing this Lt/Rt to obtain four initial channel L,R,C,S. Is correct? Which tools permit to do this? The decoding of the L,R,C,S channels from the Lt/Rt is taken care of by any Dolby Pro Logic or Pro Logic II decoder, built into most home theatre systems. This is also available in many software packages such as PowerDVD, WinDVD, etc.