View Full Version : Downmixing with custom (DPII) matrix on Mac OS X?
elmimmo
2nd April 2006, 17:21
Does anybody know of a (free-ish?) tool for Mac OS X that lets you specify a custom downmix matrix from 6 mono sources into a 1 stereo output?
I am trying to transcode Dolby Digital AC3 5.1 to Dolby Pro Logic II on a Mac, not necessarily realtime (since this is for DVD ripping anyway). I know mAC3dec can downmix to a fixed Dolby Pro Logic (non-II) stereo file, but that is not enough. I know how to extract the AC3 track from a DVD, and how to split it into 6 mono sources, but am completely lost at how to downmix it to a stereo file using a custom downmix matrix, such as the ones specified in this thread (http://forum.doom9.org/showthread.php?t=105869)?
Yes, not the first time I ask this, but thought about giving it a dedicated thread, just in case someone missed it.
specise_8472
2nd April 2006, 22:06
Does anybody know of a (free-ish?) tool for Mac OS X that lets you specify a custom downmix matrix from 6 mono sources into a 1 stereo output?
I am trying to transcode Dolby Digital AC3 5.1 to Dolby Pro Logic II on a Mac, not necessarily realtime (since this is for DVD ripping anyway). I know mAC3dec can downmix to a fixed Dolby Pro Logic (non-II) stereo file, but that is not enough. I know how to extract the AC3 track from a DVD, and how to split it into 6 mono sources, but am completely lost at how to downmix it to a stereo file using a custom downmix matrix, such as the ones specified in this thread (http://forum.doom9.org/showthread.php?t=105869)?
Yes, not the first time I ask this, but thought about giving it a dedicated thread, just in case someone missed it.
The downmix matrix (s) that are mentioned sound okay, but are not true DPLII. It involves very precise mixing to get a good balance of sound. You can absolutly hear the difference when comparing these matrix to the Surcode DPLII encoder that has just been released. I recommend you or anyone to go out and purchase it if you want proper DPLII mixing.
elmimmo
3rd April 2006, 21:19
I recommend you or anyone to go out and purchase it [Surcode DPLII encoder] if you want proper DPLII mixing.
Did you just go and recommend a $495 software for home DVD ripping activities or what? ^_-
Anyhow, back to the point. How do I check in OS X how okey those download matrix are? How do I downmix to stereo using custom matrix in OS X?
3dsnar
5th April 2006, 08:33
The downmix matrix (s) that are mentioned sound okay, but are not true DPLII. It involves very precise mixing to get a good balance of sound. You can absolutly hear the difference when comparing these matrix to the Surcode DPLII encoder that has just been released. I recommend you or anyone to go out and purchase it if you want proper DPLII mixing.
Actually it seems that these equations work properly
http://forum.doom9.org/showthread.php?s=&threadid=57988
The incorect decoding is related often to the fact, that many
software decoders simply do not work as they should.
specise_8472
5th April 2006, 18:08
Actually it seems that these equations work properly
http://forum.doom9.org/showthread.php?s=&threadid=57988
The incorect decoding is related often to the fact, that many
software decoders simply do not work as they should.
I know it sounds good, but I find it hard to beleive that it is true DPLII as true mix requires phase shifted 90 degree rear channels. This is from all the reading I have done and from the Patents. Actually from one of the latest Patents, it is revealed that DPLII is based around the Quad QS format.
scharfis_brain
5th April 2006, 18:32
the 90° phase shift IMO is nonsens, because 90° phase shift is only valid for a single frequency.
with other frequencies the phase shift will deviate.
specise_8472
5th April 2006, 18:43
the 90° phase shift IMO is nonsens, because 90° phase shift is only valid for a single frequency.
with other frequencies the phase shift will deviate.
Then you had better tell Dolby Laboratories that they have got it wrong all these years.
scharfis_brain
5th April 2006, 19:12
the problem is that I can read everywhere, that DPL should be +90° and -90° mixing. but how does one establish a filter of a constant 90° phase shift through the whole audible spectrum?
IMO this +90°/-90° shifter thing is nothing than a very small delay, that is followed by a 0°/180° shifted encoding...
But you may proove me wrong with patent papers and technical papers for DPL(2) or an explantation of that issue ;)
specise_8472
5th April 2006, 23:19
the problem is that I can read everywhere, that DPL should be +90° and -90° mixing. but how does one establish a filter of a constant 90° phase shift through the whole audible spectrum?
IMO this +90°/-90° shifter thing is nothing than a very small delay, that is followed by a 0°/180° shifted encoding...
But you may proove me wrong with patent papers and technical papers for DPL(2) or an explantation of that issue ;)
You are right. 90degree phase shift is the Hilbert Transform. Plenty on the net a Google search will turn up:)
But a poor mans DPL is as you say 180 with delay..
Try this for an excellent explanation of this implementation.
http://www.gamasutra.com/features/sound_and_music/19981218/surround_03.htm
As you can see, DPLII encoding still involves more than just mixing the signals with a set of co-efficients.
specise_8472
5th April 2006, 23:23
the problem is that I can read everywhere, that DPL should be +90° and -90° mixing. but how does one establish a filter of a constant 90° phase shift through the whole audible spectrum?
IMO this +90°/-90° shifter thing is nothing than a very small delay, that is followed by a 0°/180° shifted encoding...
But you may proove me wrong with patent papers and technical papers for DPL(2) or an explantation of that issue ;)
Also another good use of this is take the Stereo signal, Invert it (NEG), delay it (~20 samples for 44.1), swap channels, then mix back into the original signal.
This creates a reasonably good crosstalk canceller.
scharfis_brain
6th April 2006, 06:08
Also another good use of this is take the Stereo signal, Invert it (NEG), delay it (~20 samples for 44.1), swap channels, then mix back into the original signal.
This creates a reasonably good crosstalk canceller.
Such a thing my dad built about 25 years ago. It was a small little box with a linear control over Mono...Stereo...Mono-Cancellation.
THis is called "Basisbreitenregler" in German. And was used to make sound the Stereo image wider.
elmimmo
2nd May 2006, 22:51
Er… yeah, nice chit-chat. ^_-
What about regarding those stereo downmixing tools for Mac OS X, or something Linux-ish and command-line that might compile on it?
vBulletin® v3.8.5, Copyright ©2000-2012, Jelsoft Enterprises Ltd.