For a very large public listening space I am looking at a forced stereo image since only a few of the listeners will be in the sweet spot where 2 channel has a good phantom image.
I would like to use 5 speakers where the center speaker only plays what L and R have in common, the middle pair plays L or R and the outer pair has the content unique to L or R boosted by a adjustable amount.
I looked at my miniDSP flex 8 and don't think it can do this. Are there existing analog or digital solutions that do this? Has anyone tried something similar? I see some older info on doing it with 3 channels in stead of 5 and often the center channel is just L an R summed instead of only the content common to L&R
I would like to use 5 speakers where the center speaker only plays what L and R have in common, the middle pair plays L or R and the outer pair has the content unique to L or R boosted by a adjustable amount.
I looked at my miniDSP flex 8 and don't think it can do this. Are there existing analog or digital solutions that do this? Has anyone tried something similar? I see some older info on doing it with 3 channels in stead of 5 and often the center channel is just L an R summed instead of only the content common to L&R
Not sure where I should have posted this question.
Here I am bumping my own post by clarifying my question. What I want to do is:
Instead of 2 front speakers I want to set up 5 front speakers for stereo playback. The center channel is L+R summed. By moving a real or virtual slider I want to be able to pan the stereo image wider by reducing the content that is is unique to L and R to the outer pair of speakers while moving the content that is not unique to L or R to the center.
Also please let me know if you think there is a part of the forum better suited for this question.
Here I am bumping my own post by clarifying my question. What I want to do is:
Instead of 2 front speakers I want to set up 5 front speakers for stereo playback. The center channel is L+R summed. By moving a real or virtual slider I want to be able to pan the stereo image wider by reducing the content that is is unique to L and R to the outer pair of speakers while moving the content that is not unique to L or R to the center.
Also please let me know if you think there is a part of the forum better suited for this question.
I use Gerzon's method to upmix stereo 2-channel to 3-channel stereo. He outlines how to upmix to 5 channels in his paper (attached). Don't get spooked by the angle parameter - it just converts into some decimal, which is a weighting factor. There is also a "width" parameter that will reduce the stereo width towards the center.
I do the mixing using DSP in software. I don't think you will find a DSP unit (like miniDSP) that will do the upmix to 5 channels but there is likely software that has that feature. Software and computer audio I/O brings a bit of latency, so if you need this to be "real time" that you probably cannot use software. It's possible to do it using analog active circuits following Gerzon's method, which will have no latency. Implementing a real-time width adjustment is getting a little more advanced, but it is possible.
If you know your program material ahead of time, you can always perform the upmixing offline and then just play back the 5-track upmixed version during performance or playback.
I do the mixing using DSP in software. I don't think you will find a DSP unit (like miniDSP) that will do the upmix to 5 channels but there is likely software that has that feature. Software and computer audio I/O brings a bit of latency, so if you need this to be "real time" that you probably cannot use software. It's possible to do it using analog active circuits following Gerzon's method, which will have no latency. Implementing a real-time width adjustment is getting a little more advanced, but it is possible.
If you know your program material ahead of time, you can always perform the upmixing offline and then just play back the 5-track upmixed version during performance or playback.
Those papers are great. Thanks. The search for practical solutions really made my head spin. I had no idea this subject is so vast and there seems to be zero consensus. Also 98% is related to movie sound / surround sound.
After a lot of browsing I feel that maybe I shouldn't go for upmixing with a center channel. The upmixing creating the least weirdness for music playback should have an even number of speakers.
Is there a passive circuit that takes the amplifier output and uses phase inversions and summations to create 4 channels? Where the outer pair has boosted stereo content and the inner pair reduced stereo content.
After a lot of browsing I feel that maybe I shouldn't go for upmixing with a center channel. The upmixing creating the least weirdness for music playback should have an even number of speakers.
Is there a passive circuit that takes the amplifier output and uses phase inversions and summations to create 4 channels? Where the outer pair has boosted stereo content and the inner pair reduced stereo content.
What software are you using for upmixing? Does it run on a regular windows machine?
I use Gstreamer. It's not a GUI program so you have to write a sort of macro to tell it what to do and then run that from the command line. This also sets the buffer size to 1024 samples. There is evidently a Windows binary but I run it under Linux. The pipeline element that implements the mixing is called audiomixmatrix.
If you want to limit yourself to 3 stereo channels you can implement that using an analog active circuit and then there is no latency. It's really only an issue if you are also playing video and need audio and video to be in sync.
If you want to limit yourself to 3 stereo channels you can implement that using an analog active circuit and then there is no latency. It's really only an issue if you are also playing video and need audio and video to be in sync.
If you start with XLR differential signals you get the analog implementation almost for free..
For a nice DSP platform - try CamillaDSP!?
//
//
Something like this:
ok lets start with the XRL connections:
left L+ L-
right R+ R-
surround left R+ L+
surround right R- L-
center : make mono with 4 470 ohm resistors:
R+ ---470---|______ center +
L+ ---470---|
R- ---470---|______ center -
L- ---470---|
(sorry for the bad graphics 🙂 )
ok lets start with the XRL connections:
left L+ L-
right R+ R-
surround left R+ L+
surround right R- L-
center : make mono with 4 470 ohm resistors:
R+ ---470---|______ center +
L+ ---470---|
R- ---470---|______ center -
L- ---470---|
(sorry for the bad graphics 🙂 )
Is this a variation of the Hafler circuit?
I was looking at maybe getting a Chase HTS-1 which is a commercial implementation of the Hafler circuit.
Found some good info here: https://audiokarma.org/forums/index...ay-to-boost-rear-speaker-volume.792082/page-2
I am pursuing this in a developing system. The Schiit Syn is designed for this purpose and works very well for me at this point. It's main advantage is that it uses active buffers to implement the matrix components which (my guess) allows them to do some additional processing to (perhaps) limit the amount of L and R information in the derived center channel, and also have 'purer' L and R 'only' to adjust the matrixing a bit with the added controls for 'Width' and 'Presence'. I would love to see (and hopefully build) a design for the Gerzon solution.
Skip
Skip
NO, the Halfler citcuit works on the speaker outputs by putting the surrounds in series with the mains.
My proposal makes the left-right signal by taking the right pins from the XLR connectors.
But if your pre-amp/source outputs RCA you better go for the Schiit Syn as proposed above.