Ive used your file posted in post number 3
Ill do another test tomorrow without normalize checked as its getting too late tonight!
do you have phase shift when you use the convolution?
From what I could see the shuffler in post 3 was less well defined. The later posted shufflers are a bit cleaner.
Try the v2 version: http://www.diyaudio.com/forums/multi-way/277519-fixing-stereo-phantom-center-2.html#post4399454 or the first high passed version instead: http://www.diyaudio.com/forums/multi-way/277519-fixing-stereo-phantom-center-4.html#post4400043
A plot of what they do looks like this:
Looking at the plot here the high passed shuffler would probably work best.
I'll await the 900 Hz high passed version and test that.
But I like it that you could hear a difference. Surprising isn't it? My dream solution would still be to be able to only affect the mid channel. But that would take a whole lot of splitting and mixing channels to achieve.
Meanwhile I like the LoCo results as a replacement of my own mid-side processing. I liked what the JRiver DSP effect "Surround Field" did for my stage. Though it had draw backs too. Plus no way to control the effect. Black box processing. I was never able to mimic it completely with Voxengo's MSED.
This "S" curve EQ on mid and side channels gives me very similar results to what I liked about the "Surround field" processing that I can actually control.
It's probably the insertion of an anti phase signal in the opposite channel that enhances the left and right channels. Working a bit like cross talk cancelation.
But combining it with the phase shuffler did not work for me yet.
Pano, wesayso
Seen somewhere pretty shure you both own a DEQ2496 and wonder is there any relation from its "WIDTH" meny to subject this thread. Attach description from DEQ2496 manual below.
Thanks sharing interesting stuff, at present not in position to test myself and help share the experience because my setup run a mono speaker feeded stereo signal, when that change will try the exercises from here.
Seen somewhere pretty shure you both own a DEQ2496 and wonder is there any relation from its "WIDTH" meny to subject this thread. Attach description from DEQ2496 manual below.
Thanks sharing interesting stuff, at present not in position to test myself and help share the experience because my setup run a mono speaker feeded stereo signal, when that change will try the exercises from here.
Attachments
Although I do own one, it's sitting in a box upstairs, never been used, bought new in 2011(!). I started with JRiver and never bothered to unpack the DEQ due to the results I was getting.
From what I read there it is indeed doing mid/side processing with adjustable results for bass and higher frequencies. I had planned on playing with it but never got around to installing the DEQ.
From what I read there it is indeed doing mid/side processing with adjustable results for bass and higher frequencies. I had planned on playing with it but never got around to installing the DEQ.
The behringer width control speaks of "shuffle" too. But lets be clear on this, it is most probably based on the Blumlein shuffle idea.
The phase shuffle is indeed a totally different concept.
I'm not complaining yet about my experiments with the LoCo idea. It allows me to get the mid and sides sound more identical in tonality. I'm not after a width improvement. The Mid/Side shuffling does give me more definition, a fuller sound for each part in the music, better definition would be a better way to say it I guess.
I first started to use it after switching from IIR filters to FIR. The FIR processing made my huge arrays sound small, especially compared to the IIR processing.
The mid/side shuffling gave the instruments and voices the body back that I was missing. But their location didn't change. Just a bit more clear and defined.
This "S" curve Mid/Side processing seems to be able to give me the tonality in the phantom center I was missing. And without making the center seem more distant.
It takes some experimentation to get right but so far I'm liking it. Especially obvious with songs with backing vocals by the same person singing the main center voice.
The phase shuffle is indeed a totally different concept.
I'm not complaining yet about my experiments with the LoCo idea. It allows me to get the mid and sides sound more identical in tonality. I'm not after a width improvement. The Mid/Side shuffling does give me more definition, a fuller sound for each part in the music, better definition would be a better way to say it I guess.
I first started to use it after switching from IIR filters to FIR. The FIR processing made my huge arrays sound small, especially compared to the IIR processing.
The mid/side shuffling gave the instruments and voices the body back that I was missing. But their location didn't change. Just a bit more clear and defined.
This "S" curve Mid/Side processing seems to be able to give me the tonality in the phantom center I was missing. And without making the center seem more distant.
It takes some experimentation to get right but so far I'm liking it. Especially obvious with songs with backing vocals by the same person singing the main center voice.
I've been working on this kind of thing, on and off, since the early 1980's. It won't leave me alone... From reading all of these posts and more, I come up with the following:
If the center phantom image is mostly weakened in the upper-midrange frequencies, by the arrival time differential at the listeners ears, due to the size of the wavelengths involved (makes perfect sense to me), then in an effort to do minimal damage to the overall stereo sound field and sense of width, we might want to roll off the center channel speaker drive below about 1kHZ (perhaps very gradually or with a shelf ?), and above about 6kHZ ? (someone pointed out that cymbals got erroneously pulled into the center, or something like that).
With this, you’ll still have some crosstalk correlation issues at the listener position (how the waves add and/or cancel), of the now 3 channels of sound, but maybe not too bad.
David Griesinger seems to have found that the trick to generating good stereo reverb (many moons ago), was to de-correlate the two outputs. Could we achieve a better sense of separation using de-correlation here? How does one de-correlate one signal relative to other signals? I'm not an expert on this, but I suspect it's by time differential. A time delay differential between center and sides, long enough to give a decent de-correlation effect, but short enough to not be perceived as a delay differential (if that's possible).
If there’s enough delay differential between the center signal and the L and/or R signals, the spatial comb filter effect cancellations should be so close together in the upper midrange frequencies that their audibility should be minimal. I could be wrong about this, but I think it’s right. This may cause an effective de-correlation (?). So you want enough delay to cause an effective de-correlation, but not so much that the center image feels pushed back or moved forward too far.
I've got an old Carver tuner/preamp that only has Dolby ProLogic 1... (so I just use it for regular stereo), but the center channel signal has a time delay knob on the front panel, that allows you to move the center out either way, in time, from the rest (L and R and surrounds). I've never actually used this center output, but now I think I will, and see what it offers (as soon as I finish my center channel speaker project – which is about half done).
I might also add an L-XR and R-XL matrix function to the L and R signals, to further cancel the L+R signal in the L and R outputs, but I'm not sure if that helps more than it screws things up. As you increase that effect, you may be introducing signals that don't add correctly off axis. I’ll have to think about that. Plus, since most recordings have mono bass, that bass would be attenuated by an L-XR function. You’d want to bandwidth limit the L-XR processing to no lower than about 100HZ.
If the center phantom image is mostly weakened in the upper-midrange frequencies, by the arrival time differential at the listeners ears, due to the size of the wavelengths involved (makes perfect sense to me), then in an effort to do minimal damage to the overall stereo sound field and sense of width, we might want to roll off the center channel speaker drive below about 1kHZ (perhaps very gradually or with a shelf ?), and above about 6kHZ ? (someone pointed out that cymbals got erroneously pulled into the center, or something like that).
With this, you’ll still have some crosstalk correlation issues at the listener position (how the waves add and/or cancel), of the now 3 channels of sound, but maybe not too bad.
David Griesinger seems to have found that the trick to generating good stereo reverb (many moons ago), was to de-correlate the two outputs. Could we achieve a better sense of separation using de-correlation here? How does one de-correlate one signal relative to other signals? I'm not an expert on this, but I suspect it's by time differential. A time delay differential between center and sides, long enough to give a decent de-correlation effect, but short enough to not be perceived as a delay differential (if that's possible).
If there’s enough delay differential between the center signal and the L and/or R signals, the spatial comb filter effect cancellations should be so close together in the upper midrange frequencies that their audibility should be minimal. I could be wrong about this, but I think it’s right. This may cause an effective de-correlation (?). So you want enough delay to cause an effective de-correlation, but not so much that the center image feels pushed back or moved forward too far.
I've got an old Carver tuner/preamp that only has Dolby ProLogic 1... (so I just use it for regular stereo), but the center channel signal has a time delay knob on the front panel, that allows you to move the center out either way, in time, from the rest (L and R and surrounds). I've never actually used this center output, but now I think I will, and see what it offers (as soon as I finish my center channel speaker project – which is about half done).
I might also add an L-XR and R-XL matrix function to the L and R signals, to further cancel the L+R signal in the L and R outputs, but I'm not sure if that helps more than it screws things up. As you increase that effect, you may be introducing signals that don't add correctly off axis. I’ll have to think about that. Plus, since most recordings have mono bass, that bass would be attenuated by an L-XR function. You’d want to bandwidth limit the L-XR processing to no lower than about 100HZ.
Last edited:
Bob,
If I have been following this discussion correctly there is no true center channel speaker involved. That is why they are using the term phantom for the center channel sound, it is only the result of the left and right channels normal filtering effects. I don't completely understand the posted phase plot that Wesayso has just shown as I would have thought that it was just a phase shifting of left or right channels so they don't correlate to each other and have comb filtering effects. It is a strange concept to me as we try so hard to have a linear phase from a speaker and here you are doing a wiggly phase response from each channel. I'm just observing the thread but learning just by reading the post and trying to have it enter my fuzzy brain cells on all of this digital manipulation.
If I have been following this discussion correctly there is no true center channel speaker involved. That is why they are using the term phantom for the center channel sound, it is only the result of the left and right channels normal filtering effects. I don't completely understand the posted phase plot that Wesayso has just shown as I would have thought that it was just a phase shifting of left or right channels so they don't correlate to each other and have comb filtering effects. It is a strange concept to me as we try so hard to have a linear phase from a speaker and here you are doing a wiggly phase response from each channel. I'm just observing the thread but learning just by reading the post and trying to have it enter my fuzzy brain cells on all of this digital manipulation.
Yes, correct. No center speaker, just the phantom image created between left and right. I have found that the center image has a darker tone than the sides. The paper I linked to early in the thread attempts to explain why this is. The paper proposes comb filtering of the two sources in the center as the cause of the tonality shift. This comb filtering usually causes important nulls at 2K and 6K, thus darkening the tone. That's the simple explanation.
To kill this comb filtering, the phase can be de-corrilated or "shuffled" between the two channels. The plot posted above gives you a good idea of what this looks like. the phase swings up and down +/- 30 degs. Now there is no comb filtering between the two sides. Because they sum to zero, there is no frequency response change to either side, just elimination of the comb filter. That kills the notches and the tonal shift.
Yes, you have to be willing to accept this phase shuffling, normally limited to above 700-900Hz, but so far we are finding that it doesn't damage the image. Oddly enough. It just changes the tone.
I'm trying to get a number of listeners with different systems to tell us what they hear, what they like or don't like about it.
To kill this comb filtering, the phase can be de-corrilated or "shuffled" between the two channels. The plot posted above gives you a good idea of what this looks like. the phase swings up and down +/- 30 degs. Now there is no comb filtering between the two sides. Because they sum to zero, there is no frequency response change to either side, just elimination of the comb filter. That kills the notches and the tonal shift.
Yes, you have to be willing to accept this phase shuffling, normally limited to above 700-900Hz, but so far we are finding that it doesn't damage the image. Oddly enough. It just changes the tone.
I'm trying to get a number of listeners with different systems to tell us what they hear, what they like or don't like about it.
Hi,
I don't see any oddity here, just natural phenomena. Sound localisation above about 1 kHz is dominated by interaural level difference, that is the amplitude difference between the ear signals. Phase loses its importance at high frequencies.
There may be other solutions too. Some say phase randomisation is achieved by bending wave transducers due to their operational principle, without additional phase shifters.
Or if one allows some early room reflections, then cross talk comb notches are filled and become non perceivable. Thus speaker directional patterns becomes important.
One could say highly directional speakers propably benefit more of "phase shuffler" than low directional speakers.
.
If you manage to incorporate that phase response included into a passive cross-over, then this method can become very popular
.
That would be interesting to test. Never heard that before.
Quite so, yes. Much of this is covered in the white paper.
Hmmm
Tall order. Gotta think about that one.
POS got me thinning about rePhase so I had a go at making a phase shuffler with it. Made two versions, which you can see below. It's a very different impulse from the original file I did. This affects phase only, not amplitude, as the original shuffler did. Certainly sounds different. The original is a series of quick reflections 11 samples apart that fall off by 5 or 6dB each. The new ones don't work that way at all, it's purely phase manipulation.
If you get a chance, check out the new ones and tell me what you hear.
Also included in the zip file is a perfect impulse. If you run this in your convolver, you should hear no change in the sound at all. If you do, there is something wrong. You can test if convolution is really working by loading the Telephone BW file, you will certainly hear that one if it's working! It is there just to verify that your convolution engine is working. No mistaking the effect if it works.
For those of you who can't use convolution in your playback chain, I will post some files next week that have been preprocessed.
Have fun with the new versions, let me know what you hear, if anything. Cheers!
If you get a chance, check out the new ones and tell me what you hear.
Also included in the zip file is a perfect impulse. If you run this in your convolver, you should hear no change in the sound at all. If you do, there is something wrong. You can test if convolution is really working by loading the Telephone BW file, you will certainly hear that one if it's working! It is there just to verify that your convolution engine is working. No mistaking the effect if it works.
For those of you who can't use convolution in your playback chain, I will post some files next week that have been preprocessed.
Have fun with the new versions, let me know what you hear, if anything. Cheers!
Attachments
Briefly listened to the first rephrase shuffler. This one didn't move the position of the sound on the sides for me (that's a good thing). Though it still did seem to push the center image away a bit. I didn't have both tracks (one with convolved shuffler and the original) at the same volume and had to adjust on the fly. They should be exactly the same to form an honest opinion.
I'm testing too many things at once a.t.m. to say anything really useful about the coherency.
I just might have found the improvement I was after in the mid/side processing described in the LoCo link a few pages back. That's still keeping me happy so far and was simple to incorporate because I had all components in place.
I do hope more will listen to this one. If anything I think in my setup the old shuffler was more effective to make a perceivable (positive) change in the tonality of the phantom center. I'll have to give it another go soon.
I'm testing too many things at once a.t.m. to say anything really useful about the coherency.
I just might have found the improvement I was after in the mid/side processing described in the LoCo link a few pages back. That's still keeping me happy so far and was simple to incorporate because I had all components in place.
I do hope more will listen to this one. If anything I think in my setup the old shuffler was more effective to make a perceivable (positive) change in the tonality of the phantom center. I'll have to give it another go soon.