Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Dolby Headphone Wrapper DSP (Read 336369 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Dolby Headphone Wrapper DSP

Reply #75
Thanks, that worked.

PowerDVD also has an oldernewer version (1.10.0.270) of the "DolbyHph.dll". That version also produces clipping. So the version cannot be the cause of the problem.

Then I replaced the "Convert stereo to 4 channels" plugin by the "Dolby Pro Logic II" plugin (volume set to 100%). There is absolutely no clipping but it sounds very different.

Dolby Headphone Wrapper DSP

Reply #76
OK, but have in mind that 1.10.0.270 is *not* older than your 1.02.0.222  . This is a tricky point about how version numbers are represented sometimes. Look at file Properties, Version, File Version on both files for a clearer (and more comparable) representation.

Dolby Headphone Wrapper DSP

Reply #77
Oh, you're right. Stupid me.

Dolby Headphone Wrapper DSP

Reply #78
Hancoque, here you have a test version with checkbox for enabling/disabling Dynamic compression. Note that until now it was always in enabled state, like did PowerDVD and WinDVD.

Test version

Dolby Headphone Wrapper DSP

Reply #79
Chungalin, could you please elaborate on what that fix does please?

edit: Hancoque, I don't know if you noticed but you're losing a bit of dynamic range when you turn the plugin volume down to 50%. On that green wave, the highs are a bit lower, and the lows are a bit higher. Hmm...

edit2: Ok, I figured it out. Replaygain is the trick. I converted (from FLAC) two files: the first was just converted to 4 channels and then Dolby Headphones. Cooledit Pro reports 13990 and 13537 possibly clipped samples in the left and right channels, respectively. The other file was also 4 channels and Dolby Headphones, BUT I added replaygain (album mode). This file has 0 clipped samples (same as the original file). In the Dolby Headphone config, the volume was set to 100% at all times. This leads me to believe that Foobar applies RG then applies DSP. This might be common knowledge but it's the first I've heard of it. Anyway, since you're obviously having clipping problems, there is your solution

Dolby Headphone Wrapper DSP

Reply #80
@Chungalin: I will check it out later today.

@hushypushy: I think that will only help if the material's volume is reduced by at least 6 dB. Otherwise there could still be clipping. And by the way: Wouldn't applying ReplayGain have the same impact on dynamic range as lowering the volume in the DSP chain?

Dolby Headphone Wrapper DSP

Reply #81
This version is not a fix, it just shows a new control in configuration panel that allows to enable/disable a parameter inside DH processor that I still don't know precisely what it does. It seems to affect volume and once I noticed that when it wasn't enabled it produced samples out of range (while in float mode). Under Foobar I set float mode with a range of 1.000 (-1...+1). Under PowerDVD I saw that it did set a range of 0.99999. Could this be significative?

Dolby Headphone Wrapper DSP

Reply #82
It produces some weird sound, but it doesn't matter... i don't use it

Dolby Headphone Wrapper DSP

Reply #83
@hushypushy: I think that will only help if the material's volume is reduced by at least 6 dB. Otherwise there could still be clipping. And by the way: Wouldn't applying ReplayGain have the same impact on dynamic range as lowering the volume in the DSP chain?


Hmm, this is true, but then the same goes for any software volume changes, yeah? Optimally, you'd want to keep the software at 100% everywhere and change the volume with your speakers.

Anyway, RG offers customized gain changes, so each file will be affected differently with regards to dynamic range, whereas setting the internal volume setting will affect each file in a static way, some worse than others. I suppose that the files that are closest to 89dB (smallest RG value, either positive or negative) will remain the most faithful to the original files?

One last thing, from http://replaygain.hydrogenaudio.org/faq_noise.html:

Quote
But in reality, a consistent replay level (an advantage which EVERYONE can appreciate) will outweigh the problems of digitally scaling the data (which is a problem that 0.01% of listeners are going to notice).

Dolby Headphone Wrapper DSP

Reply #84
I have made a new graphical comparison. This time not only between 50% and 100% volume but also between "Clipping" activated and deactivated.



As you can see, clipping occured only with "Clipping" deactivated and 100% volume. But if "Clipping" in fact enables dynamic compression, wouldn't it be better to deactivate it?

Dolby Headphone Wrapper DSP

Reply #85
But if "Clipping" in fact enables dynamic compression, wouldn't it be better to deactivate it?


Aren't you asking to choose between clipping and dynamic range? 

With consideration, it seems that it's okay to lose a bit of dynamic range (whether it be via the plugin or via RG or whatever) than to gain a ton of clipping (as both of our tests show happens).

I want to try it with that new PLII DSP but I can't find the DLL

edit: Just did a couple of tests with PLII and DHP...I like the stereo to 4 channel --> DHP much more than PLII --> DHP. Although, having spent way too much time listening to music on a 5.1 system with PLII, I can see why I like the PLII to headphone effect much less than just a doubling of the stereo sound, so let's see if I can explain it...well, using the 4 channel converter, it just makes a big sound field, and listening to a stereo source like this gives me the impression that I'm sitting directly between two large speakers. Using PLII though, it just doesn't feel...right. It feels kind of cheap, and tinny. When I listen to PLII on 5.1, I really enjoy it because it preserves good front stereo sound, moving some things like vocals and solos to the center, and moving extra tidbits or layers to the back. What I'm trying to say is, for my tastes, PLII is much better suited to real 5.1, whereas Dolby Headphone is useful for creating an "out of head" sound field which, IMO, works best with dual stereo.

Dolby Headphone Wrapper DSP

Reply #86
Quote
edit: Just did a couple of tests with PLII and DHP...I like the stereo to 4 channel --> DHP much more than PLII --> DHP. Although, having spent way too much time listening to music on a 5.1 system with PLII, I can see why I like the PLII to headphone effect much less than just a doubling of the stereo sound, so let's see if I can explain it...well, using the 4 channel converter, it just makes a big sound field, and listening to a stereo source like this gives me the impression that I'm sitting directly between two large speakers. Using PLII though, it just doesn't feel...right. It feels kind of cheap, and tinny. When I listen to PLII on 5.1, I really enjoy it because it preserves good front stereo sound, moving some things like vocals and solos to the center, and moving extra tidbits or layers to the back. What I'm trying to say is, for my tastes, PLII is much better suited to real 5.1, whereas Dolby Headphone is useful for creating an "out of head" sound field which, IMO, works best with dual stereo.


So you successfully got me into this thread, hushypushy
Well, I see two possible reasons for this:
1. 4ch mixer is louder, especially the bass since that gets played on 4 virtual speakers instead of 2 (the front speakers) as with  DPL2. Maybe one should compare both modes with replaygain enabled (e.g. by converting a track 2 times, once with each dsp settings).

2. Maybe you don't expect to get "real" surround with Dolby Headphone, since you don't know where the speakers are. Actually it can do much more than just "out-of-head", because it can produce things like "left behind you". Have you played the channeltest.mp3 (on my site)? For me its sounds very much like the recommended home theatre setup with rear speakers about 10-20 degrees behind you.

Given these 2 things it comes down to this (in my eyes):
Do you like the same music more if you get 2 times the bass?
If I want more bass or more volume, I turn it up. But if I want to hear the audience shouting behind my perceived back, I cannot get around PL2 (or anything like it).

So please try it again!
Btw: I had the surround channels at 75% volume because for my taste there was way too much action in the rear speakers. Didn't expect that it could be vice versa. Just changed this, so you might redownload if you wish.

Have to go to work now!

Christian

Dolby Headphone Wrapper DSP

Reply #87
Aren't you asking to choose between clipping and dynamic range? 

With consideration, it seems that it's okay to lose a bit of dynamic range (whether it be via the plugin or via RG or whatever) than to gain a ton of clipping (as both of our tests show happens).
I think that lowering the volume by 6 dB (to 50%) is okay if you can get rid of any compression (and clipping) in exchange. If processing takes place in 32-bit Float then it shouldn't really be an issue.

edit: Just did a couple of tests with PLII and DHP...I like the stereo to 4 channel --> DHP much more than PLII --> DHP.
Yes, I can second that. While the PL2 plugin produces a wider soundstage the sound also becomes kind of tinny. The PL2 plugin may be a superb plugin to get stereo sound to a real 5.1 system (which I can't test because I only have 2 speakers) but in combination with the DHP plugin it just doesn't feel right. Maybe it's a question of settings in the PL2 plugin (to which users currently have no access).

2. Maybe you don't expect to get "real" surround with Dolby Headphone, since you don't know where the speakers are. Actually it can do much more than just "out-of-head", because it can produce things like "left behind you".
Can it really? After all you only have two channels as input (2.0 -> 5.1 with the PL2 plugin) and two channels as output (5.1 -> 2.0 with the DHP plugin). How can you have something "behind you" if there is neither "behind"-information in the source material nor any physical "behind"-information in the two output channels?

Btw: I had the surround channels at 75% volume because for my taste there was way too much action in the rear speakers. Didn't expect that it could be vice versa. Just changed this, so you might redownload if you wish.
It would be great if you could make as most settings available to the user as possible. That way one would be able to tweak the plugin for an optimal use. After all I believe that a professional upmixing algorithm like Dolby Pro Logic II has to have some advantages over mere channel doubling. But at the current state of the plugin I cannot say that I really like the sound it produces.

Dolby Headphone Wrapper DSP

Reply #88
Thanks Hancoque for taking your time to make these illustrative graphics. Let's analyze them carefully.

First of all, I've labeled the effect as "Clipping" because I didn't know exactly what it did. But now we can clearly see a non-linear behaviour in the blue wave that reveals a dynamic compression at work. The peaks are flattened as they get close to the "danger zone", which becomes a highly non-linear zone. We must not confuse this with plain clipping that is just (simplifying) "if (x > boundary) then x = boundary". I've changed the checkbox label to "Dynamic compression" in order to avoid confussion.

The problem with any downmixing from many channels to a single one is the balance between clipping and attenuation. The most elementary downmix formula:

(Ch 1 + Ch 2 + ... Ch n) / n

yields a result that never clips but has poor dynamic range unless all channels are fully utilized. If you reduce the divisor then volume rises, but somewhere-sometime you can get out of boundaries. The sum (mix) of n channels each one of m bits requires m+log2(n) bits in order to preserve all the information.

It has some sense that Dolby wanted a compromise between the two options adding dynamic compression. Now, what do you prefer, clipping or non-linear fuller sound?

Personally, I don't like non-linear processes for audio. From my point of view, my aim will be: upmix stereo to 4 or 5 channels in a predictable way, try to keep away of dynamic compression and set an amplification level that doesn't clip in loudest passages.

Dolby Headphone Wrapper DSP

Reply #89
So you successfully got me into this thread, hushypushy
Well, I see two possible reasons for this:
1. 4ch mixer is louder, especially the bass since that gets played on 4 virtual speakers instead of 2 (the front speakers) as with  DPL2. Maybe one should compare both modes with replaygain enabled (e.g. by converting a track 2 times, once with each dsp settings).

2. Maybe you don't expect to get "real" surround with Dolby Headphone, since you don't know where the speakers are. Actually it can do much more than just "out-of-head", because it can produce things like "left behind you". Have you played the channeltest.mp3 (on my site)? For me its sounds very much like the recommended home theatre setup with rear speakers about 10-20 degrees behind you.

Given these 2 things it comes down to this (in my eyes):
Do you like the same music more if you get 2 times the bass?
If I want more bass or more volume, I turn it up. But if I want to hear the audience shouting behind my perceived back, I cannot get around PL2 (or anything like it).


I've tested a bunch of songs up till now, and PL2 has won none of them. 4 chan -> DHP has won most, and no DSP has won a few. It's not up to loudness or extra bass, I can assure you. But I also made a test where I RG gained a FLAC file, then outputted each file with the different DSP, then RG'd those, and then tested them. I'm not disappointed in the surround aspect or the soundstage---it reminds me a bit of listening to my 5.1 system. However, it feels like I have my ears in tubes and there are speakers in the end of those; the music has this weird tinny feeling to it, I'll post samples if this doesn't make sense.

I tried the channeltest, and it reproduced surround. I saw the fronts in the fronts and the rears on the side/rear of me, which is the optimum surround setup, yeah? It seems to be a bit more evidence that both PL2 and DHP work (by doing 2.0-->5.1 then back to 2.0), but unfortunately, quite a bit of fidelity is lost.

I don't have 5.1 speakers on my computer, so I can't test out PL2 in its proper form. But I'm wondering, why the lack of configuration? I'm guessing that it just hasn't been implemented yet (or is perhaps impossible). In cheaper amps, there is at least the "Game", "Movie", and "Music" mode, and in nicer amps they actually let you set the parameters. Is there any way that can be implemented in this wrapper? A feature request if I could--a dialog box that simply asks you to select where the DLL file exists.

Quote
Can it really? After all you only have two channels as input (2.0 -> 5.1 with the PL2 plugin) and two channels as output (5.1 -> 2.0 with the DHP plugin). How can you have something "behind you" if there is neither "behind"-information in the source material nor any physical "behind"-information in the two output channels?


You missed an important note--when it does convert to 5.1, there is some rear speaker info, and then DHP processes that. It's a subtle effect though (most of the time), and from what I've read, that's what they were going for when they created it. So in that aspect, it sometimes falls short, it is a matrix after all. Listening to, for example, Dark Side of the Moon not a good idea (especially when quad/5.1 versions exist), but listening to 10,000 Days is excellent.

Dolby Headphone Wrapper DSP

Reply #90
Quote
Can it really? After all you only have two channels as input (2.0 -> 5.1 with the PL2 plugin) and two channels as output (5.1 -> 2.0 with the DHP plugin). How can you have something "behind you" if there is neither "behind"-information in the source material nor any physical "behind"-information in the two output channels?


Yes, it can. If the source is encoded in Dolby Stereo (which exist since ~1970).
For each sound source you determine where it is, relative to the listener, along 2 major axes:
left <-> right and center <-> rear.  Then, it's encoded like this (taken from wikipedia):
Left Total  = Left*1.000 + Right*0.000 + Center*0.707 + RearLeft*j*0.8165 + RearRight*j*0.5774
Right Total  = Left*0.000 + Right*1.000 + Center*0.707 + RearLeft*k*0.5774 + RearRight*k*0.8165
j = + 90º phase-shift , k = - 90º phase-shift

Its hard to get the original directions back from Left Total / Right Total, but its possible (with some assumptions such as relative stability of locations/frequencies over time and limited number of separable sources - probably up to 4).

This is your "behind" information in the source.

So where is the behind information in the output? Well, how can you determine if anything is behind you if your brain gets only two streams of audio (left ear+right ear)?
It can do this *only* because the sound coming from behind you sounds differently from sound coming from infront of you. That works because sound either gets blocked or reflected by your earlobes.
Cut them off and you loose your spatial information. This difference can be recorded and reproduced electronically. Listen to the samples at http://binaural.jimtreats.com/

This is your "behind" information in the output.


Quote
It would be great if you could make as most settings available to the user as possible. That way one would be able to tweak the plugin for an optimal use. After all I believe that a professional upmixing algorithm like Dolby Pro Logic II has to have some advantages over mere channel doubling. But at the current state of the plugin I cannot say that I really like the sound it produces.


Yeah, I'd like to add a separate volume control for each channel (I think this would be the most important thing - if no other plugin like channel mixer can do this), but doing all this GUI c*** is what I hate most about software development and I need a bit free time for that.
The second possible option is "music mode" versus "movie mode". Its in music mode by default, b/c movie mode makes not much sense within a music player, I think.
And I believe that this implementation doesn't have any more advanced options like "center width" and "dimension" like some more expensive A/V receivers have.
If you find any software player that supports this in its PL2 implementation then tell me please (I'm interested).

Quote
I've tested a bunch of songs up till now, and PL2 has won none of them. 4 chan -> DHP has won most, and no DSP has won a few. It's not up to loudness or extra bass, I can assure you. But I also made a test where I RG gained a FLAC file, then outputted each file with the different DSP, then RG'd those, and then tested them. I'm not disappointed in the surround aspect or the soundstage---it reminds me a bit of listening to my 5.1 system. However, it feels like I have my ears in tubes and there are speakers in the end of those; the music has this weird tinny feeling to it, I'll post samples if this doesn't make sense.


I believe you, although I cannot second that in general. But I'd definitely like to hear a few of those samples.
I found some stuff that sounds equally good, e.g. Dover, Coldplay, some others.
And some cinematic soundtracks like Once Upon A Time In Mexico (especially) or Gladiator seem to sound better with DPL2.

But I think I have understood what you mean by tinny. Since I use PL2 in exactly the same way as the Program that ships the DLL I suspect that this is an inherent problem of the combination of DPL2 and the virtual speakers present in Dolby Headphone. I am confident that I can solve that by equalization, though.

Christian

Dolby Headphone Wrapper DSP

Reply #91
Okay, I think you're right about the "behind" information. But how often do you encounter music that contains an embedded surround signal? Isn't most music on a CD saved in plain stereo?

Dolby Headphone Wrapper DSP

Reply #92
Quote
Okay, I think you're right about the "behind" information. But how often do you encounter music that contains an embedded surround signal? Isn't most music on a CD saved in plain stereo?

That's hard to tell. Depends mostly on the recording studio/engineers, I would say.
Since you don't loose anything by encoding to Dolby Stereo (except that the surround channels cancel each other out when played on a mono device), some studios add it for free. E.g. some of my Uncle's songs (a Berlin Indie Artist) were recorded with explicit surround information and he didn't even know.
A few Artists/Bands which I know of who produced surround recordings include Jeff Beck, Nine Inch Nails, Andrea Parker, Autechre, Robert Plant, Will Smith, Shakira, Fear Factory, Queens Of The Stone Age, Dover, Cog, Rage Against The Machine, David Bowie, Skinny Puppy, Failure,  Incubus, Tool, Tito & Tarantula, ....
I just confirmed that list by putting PL2 in the Movie mode (which means something like "strict with no guessing") and turning off the front speakers.

And of course most movie Soundtracks.

Dolby Headphone Wrapper DSP

Reply #93
Whoa whoa whoa, I'd love to see some source information on that, because I've never heard that any albums were mastered with embedded surround. Like NIN, QOTSA, RATM, Tool made albums that have matrixed surround information? Are you just saying that because they end up with a lot of rear info? Or because somewhere actually says that they were mastered that way?

edit: here is a quick sample of a song that I think sounds great with 4chan-->DHP and crappy with PL2-->DHP. It's the first 30 seconds of Jambi by Tool (I can only UL 30 seconds, right?). Especially listen to the vocals, but also how it sounds so airy and tinny. The file was first ReplayGained at -7.27dB, then each one was put through a DSP chain consisting of either 4chan-->DHP or PL2-->DHP only.

http://hushy.flaretech.net/music/jambi.rar

Dolby Headphone Wrapper DSP

Reply #94
Whoa whoa whoa, I'd love to see some source information on that, because I've never heard that any albums were mastered with embedded surround. Like NIN, QOTSA, RATM, Tool made albums that have matrixed surround information? Are you just saying that because they end up with a lot of rear info? Or because somewhere actually says that they were mastered that way?


You guessed right, the source of this info is my harddisk. So, if I had a digital camera I could send you a pic of my source, but I don't ;-)
Ok, all jokes aside, I fully understand your suspiciousness. I called a few names just based on the fact that some of their albums contain a lot of "action" in the surround channels. I guess I was a bit quick with that, so I spent much time today just to find out that a lot of this is just crosstalk or sound leakage.
In plain stereo recordings you typically get a garbled, bandpass-filtered copy of the fronts in the surround channels.
Good examples of this are e.g. Subway To Sally - nearly every album, Lord Of The Dance, Led Zeppelin, The Doors, Blind Guardian, John Lee Hooker, Joe Satriani, Vanessa Mae, Sean Paul - Get Busy etc.

But there are some counter-examples. Let's start with mainstream music. Take for example Alicia Keys - The Diary Of Alicia Keys. In about 1/3rd of the tracks you get
the entire background chorus in the rear speakers. While not much of the rest of the song is audible there (below 10-20% of the front volume). Its like some kind of
duet between her (mostly in the center) and the background singers (in the surrounds, but also L/R). So here it's obvious that someone put part of the song into the surround channel.
The same pattern can be seen with Black Eyed Peas - Elephunk.
Then there are cases where the entire song (including lead singer) is replicated at slightly lower volume into the surrounds, e.g. Christina Aguilera - The Voice Within. The surround actually morphs from just echo to
a full front copy over the duration of the song.
You sometimes find multiple echoes of the main voice in the surrounds while in other cases you have just the voice, only at a lower volume.
Compare e.g. the first half of the aforementioned song by Christina Aguilera (echo) to Natural - What if (copy of the lead voice at lower level)
and to BEP feat. Justin Timberlake - Where is the love (voice nearly inaudible).
So that echo effect is most likely not an artifact of DPL2 or of some natural phase effects of echoes but deliberately applied. Although I was quite surprised about how easily DPL2 gets confused and puts something into the rear channel. 2 independently positioned front sources seem to be enough under some circumstances.
Rage Agains The Machine - Renegades is a example of an album with most of the guitars in the surrounds.
(e.g. Kick Out The Jams) And the interesting thing is that they are centered. So they wouldn't be there if they weren't phase-inverted, because the basic surround info comes from Left - Right (which would cancel each other out). And I think it's highly unlikely that someone accidentally inverts the polarity of exactly one channel (either left or right) of the guitars. Now compare this with the life recordings on the same album.
In these, you have generally a much lower surround level and the surround is mostly a mixture of the other channels (again like the garbled copy that I mentioned before).
Probably an even better example is Fear Factory - Remanufacture. Here it's again mostly guitars, industrial sounds and ambient effects that end up in the surround channel. And they are louder than the fronts. So it can't be crosstalk by definition.
Next example: Snake River Conspiracy - Sonic Jihad. Here the surrounds sometimes play completely complementary stuff to what the front speakers do.
In Chemical Brothers - Push The Button you find some front/back fading, echoes, various ambient sounds, but again its not 100% clear that this is not just DPL2 going mad (but its unlikely).
Though, I was wrong with QOTSA and Robert Plant. There it's not more than crosstalk or maybe just replicated front channels. Hope that were enough examples for now.

I created a tool for you to check that for yourself, with your headphones. It sends the music through PL2, rotates the soundfield by 90° so that you get the L,C,R channels to your left and the SL, SR channels to your right side.
Since you want to measure front/back and not center/off-center, the center speaker is attenuated by the factor 0.66.
As a last step, the stereo separation (between your left and right ear) is increased by a factor of 0.3.

If anyone finds it useful it might get embedded into a later version of the foo_dsp_pl2 plugin.
Currently it's call foo_dsp_pl2-featuretest.dll and its available on http://hosted.filefront.com/prooptimizer/
Since the output of this is basically stereo, you can chain it with the stereo->4 channel upmix and DH to get more volume or less fatigue...

Christian

Dolby Headphone Wrapper DSP

Reply #95
Very interesting...I will mess with this.

By the way, a good album I like to hear on my amp + "real" PL2 is Tool - 10,000 Days. Wings for Marie/10,000 Days has a whole lot going on in the back.

Dolby Headphone Wrapper DSP

Reply #96
Hi folks,

I was just thinking about how PL2 works when it struck me that it cannot really work. At least not with music.
The problem with Dolby Surround is in how the information is encoded:
Most of us know that PL2 uses the relative phase information of the signals.
If you hate maths or geometry you can directly jump to the paragraph starting with BUT ;-)

Whether it does this by calculating L-R and L+R and comparing that somehow or by calculating the Fourier Transform of L and R plays no role:
For each time window consisting of n samples you cannot get more than n/2 amplitudes and phases angles (one for each frequency component).
So you get this info both for Left Total and Right Total.
Now you can (for each frequency component)
  a) compare the amplitudes between L and R, i.e. whether the signal is rather left or right.
  b) compare the phase angles between L and R
    0 degrees angular difference would mean that the signal is completely in front.
    180 degrees angular difference (i.e. the flipped wave) means that the signal is completely behind.
    Anything in between will be somewhere in between (front/back-wise)
    Compare this with the PL2 mixing matrix (bottom of http://en.wikipedia.org/wiki/Dolby_Pro_Logic)
   
So for each of the frequency components (say 512) in your time window you get these 2 numbers which can each be put into the range (-1 to +1).
Then you can plot these 512 points into a 2-dimensional plane and directly see where the sound sources should be located - relative to the listener
which is at (0,0).
         
If you colored the dots according to their frequency you have not lost a single bit of information, so you can fully reconstruct the original signal from this representation.
But you can also send each dot (or better its corresponding sine wave) to the nearest speakers and get surround.
Note that this is just the mathematical mapping of the surround information (as specified by Dolby) into your soundfield and is independent of what
algorithms are applied, whether one uses analog filters, negative feedback, feed forward, FFT or whatever.
If you have enough CPU horsepower you can also use this algorithm directly but you'd have to do a lot to improve stability and mitigate artifacts,
e.g. cluster the sources, track them over time, look at what frequencies occur together (and assume they belong to the same source), look at harmonics etc.


BUT now see what happens if we have 2 different sound sources within the same frequency range, e.g. a girl singing on L and another one on R.
We cannot make any assumptions about the relative phases in the girls' frequencies. Most likely there will be a phase difference
somewhere between 0 and 180 degrees between L and R.
And this means that PL2 will place it somewhere between front and back -- arbitrarily. It can not distinguish between 2 sources in front you
or a single one behind you. No algorithm in the world could do that if the girls started signing at the same time.
This implies that you cannot have a stereo choir in front of you in PL2. It will get scattered somewhere between front and rear.
The same goes for stereo guitars, especially distorted ones, etc. You name it.

So the conclusion is that applying PL2 to music is very very unpredictable and most of what you hear in the surrounds is nothing but artifacts.
Of couse it is still possible to put a source anywhere in the soundfield, but only as long as it doesn't conflict with the rest of the music.

I have to take back most of what I wrote in the previous post, since this explains almost every effect that I noticed there (including the echoes) -- stupid me!
While it is still possible that some of the surround effects one hears there were applied deliberately, it is actually much less likely that I thought.
But I'm glad that I now have the exact answer to why one should buy 5.1 recordings or why the Fraunhofer develops an mp3surround format ;-)

Christian

Dolby Headphone Wrapper DSP

Reply #97
I've been trying to understand how PL works for ages, but I'm incredibly weak towards math.

So let me ask you this: why do you think that some albums sound fantastic with PL2 applied? And why do some sound crappy? To bring up my previous example, Tool's 10,000 Days sounds absolutely fantastic and that album does sound like it was programmed for decoding---it's magnificent. But look back to another one of their albums, Ænima, and it doesn't get much seperation in the surround channels, there isn't much of a difference between listening to stereo and between 5.1 for most of that album.

Is it sheer luck? Programming techniques? After all, some albums sound clearer than others with regards to distinction between instruments and sounds, for example.

Dolby Headphone Wrapper DSP

Reply #98
Quote
So let me ask you this: why do you think that some albums sound fantastic with PL2 applied? And why do some sound crappy? To bring up my previous example, Tool's 10,000 Days sounds absolutely fantastic and that album does sound like it was programmed for decoding---it's magnificent. But look back to another one of their albums, Ænima, and it doesn't get much seperation in the surround channels, there isn't much of a difference between listening to stereo and between 5.1 for most of that album.

Is it sheer luck? Programming techniques? After all, some albums sound clearer than others with regards to distinction between instruments and sounds, for example.


There's exactly one CD stuck in my CD player. And this is Tool - 10,000 Days, which I haven't ripped yet.

But in general, one can observe that PL2 gives reasonable results in many cases, even if no one monitored it in the studio. That's because the things that end up in the surrounds are usually those with most stereo-separation in the source. This can almost be called "natural" behaviour. So it *could* be luck.
On the other hand, you can have very bad luck. Like a guitar to the left which moves to the surrounds whenever a second guitar to the right plays something.
But if one monitors PL2 during recording and if one is willing to make tradeoffs with how it sounds in plain stereo, one can tweak a thousand things. There are simple rules that can be followed to get more predictable results. E.g. putting guitars behind you is relatively easy, percussion too, keeping the singer in the center speaker is easy etc. Trying to do the opposite is often hard and requires some tradeoffs.

edit: Just listened to Tool - 10,000 Days with PL2 (-featuretest).
There are indeed a few very nice & clear effects in it, e.g. the voice in Jambi at 1:10, the echoes at 1:44,
the thunderstorm in Wings Pt. 2 or the voices in Intension (@2:10). Seems to be more than coincidence.

Christian

Dolby Headphone Wrapper DSP

Reply #99
Wow, I am finding this plugin incredibly enjoyable!  I am listening to The Phantom of the Opera soundtrack with the convert to 4 channel and the DHP, wow, it's just eons better sounding than the plain input, I was switching it on and off, and it just sounded dead with it off.  Great work!  Just a question, what is the procedure to encode music with this for playback on an ipod or other device?  Great work guys!

Thanks,
Nathan