+1
While this isn't the fixing that this thread was all about, it would be interesting to have a reference of your two different systems.
You mentioned the Kali having a huge stage. Does it do this with the majority of recordings played back on it? Same sauce on everything?
Most likely the cosynes have a stage that varies with content?
I am just starting to follow this (old) conversation so forgive me for not reading the whole tread 🙂
Question: I assume you do this processing only on material that has a dedictated center channel and then do this processing on center and mix it with the L and R channels to emulate a center speaker?
Or are you doing these (listening) tests on stereo material?
Those tests were Stereo files with the shuffler (second version linked in that thread) applied to the left and right channel (so not applied to a dedicated center like described in the paper).
It acted much like reflections and made me aware of left and right speaker positions. That's one of the reasons we moved on towards phase shufflers. I think I've tried just about everything I could think of, including trying each variant of the shuffler (phase shuffler and original peak pattern shuffler) on phantom material only within Stereo. However we can not get back to pure phantom sounds within Stereo. It's just the mid part (L+R) sum that I tried this on, that sum does still contain left and right panned sounds. I even tried to counter the change that happens in the left and right channel in the sides channel, but it won't work that way.
Ever since this thread I have used the (more regular) mid/side EQ in my Stereo setup for enhancement/listening pleasure.
For Home Theatre I use something derived from the original shuffler and a cross talk cancelation principle. Applied to (the dedicated) center channel only, for play back on my system without a physical center speaker.
It acted much like reflections and made me aware of left and right speaker positions. That's one of the reasons we moved on towards phase shufflers. I think I've tried just about everything I could think of, including trying each variant of the shuffler (phase shuffler and original peak pattern shuffler) on phantom material only within Stereo. However we can not get back to pure phantom sounds within Stereo. It's just the mid part (L+R) sum that I tried this on, that sum does still contain left and right panned sounds. I even tried to counter the change that happens in the left and right channel in the sides channel, but it won't work that way.
Ever since this thread I have used the (more regular) mid/side EQ in my Stereo setup for enhancement/listening pleasure.
For Home Theatre I use something derived from the original shuffler and a cross talk cancelation principle. Applied to (the dedicated) center channel only, for play back on my system without a physical center speaker.
Last edited:
As said above, the shuffler has been used in our tests with stereo material to correct for the darker sound of the phantom center as compared to left and right. I believe the original intention was for L-C-R recordings where a center speaker is NOT used. The lack of a dedicated center speaker has a negative impact on dialog placed in the center. The shuffler helps correct that.
Hi Pano, Wesayso, others,
About two months ago, I finished hooking up the horn system (JBL2226 + 2380 horn and 2445J driver). Using a passive crossover between the drivers. I was in such a rush that I connected the right woofer in reverse polarity. Then applied DRC to correct for the room response. Of course, DRC also fixed the phase of the right speaker such that it ends up being close, though not exactly the same, as the left speaker.
The sound was somewhat phasey, but 'Birds on a wire' from Famous Blue Raincoat was a revelation. There was so much space and air between the various instruments on that song and they were all locked in place. All music was improved. Symphonic music and other 'depth' recordings were amazing. It felt like you were listening from the perspective of where the microphones were placed during recording. Even sitting way off to the side, the soundstage remained in place. It was truly amazing. At first, I thought it was the 2380 horn, with its perfect polars that was the cause behind this amazing imaging. But after a while, the phasiness got to me and I started looking in the measurements and found the reverse woofer polarity. So, I repeated the measurements with the correct polarity, and ran it through DRC, and... the magic was gone. It measured better (phase and crossover null), but imaged less better.
I still haven't figured out what happened there, but then I started reading this thread, having played around with Pano's phase shuffler a few years ago. Then I landed on Frankl's LoCo pages and gave that a go. I used Voxengo's MS plugins for encoding and decoding and inserted their linear phase equalizer between the two and used the S-curve for the side channel--boost around 400 Hz and dip around 1.3 kHz. That did improve imaging quite a bit.
Is that what you are doing Wesayso? When you said MS enhancement, what EQ are you using? Also, I'm not confident I'm doing the LoCo technique correctly though it does improve imaging quite a lot. Does anyone have any experience with it?
I am attaching the 'phasey' sounding but superb imaging configuration measurements. Note that these have no MS processing at all. What you see is what is playing.
About two months ago, I finished hooking up the horn system (JBL2226 + 2380 horn and 2445J driver). Using a passive crossover between the drivers. I was in such a rush that I connected the right woofer in reverse polarity. Then applied DRC to correct for the room response. Of course, DRC also fixed the phase of the right speaker such that it ends up being close, though not exactly the same, as the left speaker.
The sound was somewhat phasey, but 'Birds on a wire' from Famous Blue Raincoat was a revelation. There was so much space and air between the various instruments on that song and they were all locked in place. All music was improved. Symphonic music and other 'depth' recordings were amazing. It felt like you were listening from the perspective of where the microphones were placed during recording. Even sitting way off to the side, the soundstage remained in place. It was truly amazing. At first, I thought it was the 2380 horn, with its perfect polars that was the cause behind this amazing imaging. But after a while, the phasiness got to me and I started looking in the measurements and found the reverse woofer polarity. So, I repeated the measurements with the correct polarity, and ran it through DRC, and... the magic was gone. It measured better (phase and crossover null), but imaged less better.
I still haven't figured out what happened there, but then I started reading this thread, having played around with Pano's phase shuffler a few years ago. Then I landed on Frankl's LoCo pages and gave that a go. I used Voxengo's MS plugins for encoding and decoding and inserted their linear phase equalizer between the two and used the S-curve for the side channel--boost around 400 Hz and dip around 1.3 kHz. That did improve imaging quite a bit.
Is that what you are doing Wesayso? When you said MS enhancement, what EQ are you using? Also, I'm not confident I'm doing the LoCo technique correctly though it does improve imaging quite a lot. Does anyone have any experience with it?
I am attaching the 'phasey' sounding but superb imaging configuration measurements. Note that these have no MS processing at all. What you see is what is playing.
Attachments
Interesting. 
I don't know what the LoCo adjustment is tho.
Years ago I made a similar mistake with a 15+horn combo. I got one of the 811 horns in reverse polarity. Wow what a crazy wide soundstage. It extended past the the walls of the house and was crisp. There were even sounds coming from directly behind me that I thought were the neighbors. The phantom center was crap tho, almost not there so not really viable.
With M/S processing and EQ you can do some handy things.
I don't know what the LoCo adjustment is tho.Years ago I made a similar mistake with a 15+horn combo. I got one of the 811 horns in reverse polarity. Wow what a crazy wide soundstage. It extended past the the walls of the house and was crisp. There were even sounds coming from directly behind me that I thought were the neighbors. The phantom center was crap tho, almost not there so not really viable.
With M/S processing and EQ you can do some handy things.
Frankl's LoCo Technique:
frankl's stereo pages - LoCo
And other docs he references:
http://www.sengpielaudio.com/FrequenzabhHoerereignisrichtung.pdf
The MS EQ also helps with locking the central image in one spot. It does something to the tonal balance though, it is subtle but it is there.
frankl's stereo pages - LoCo
And other docs he references:
http://www.sengpielaudio.com/FrequenzabhHoerereignisrichtung.pdf
The MS EQ also helps with locking the central image in one spot. It does something to the tonal balance though, it is subtle but it is there.
Hi ra7,
Yes, that's where my journey with mid/side EQ started. I had noticed some interesting things happen when I used the JRiver effect Surround field. I did ark the JRiver developers what it did exactly, but they wouldn't tell me. (it was a secret)
Fast forward, I used some test tones and figured out what they were using (a boost of side info compared to the mid (or phantom) part. That's about the time Pano started this thread and member CBW posted a link to the LoCo principle (page 8 of this thread).
So I tried that as well, and combined it with what I've learned on this thread.
So I use a variation of that LoCo principle, adjusted to get tonality good/similar between the panned and phantom sounds. This thread was a great help to get me experimenting and realising what did what. I should have some numbers in my thread somewhere. It isn't easy to find the right balance, but if you do: boom! Rock solid image, excellent tonality and holographic on songs that support that. Fun for just about any song. Especially with a Haas kicker brought in.
Yes, that's where my journey with mid/side EQ started. I had noticed some interesting things happen when I used the JRiver effect Surround field. I did ark the JRiver developers what it did exactly, but they wouldn't tell me. (it was a secret)
Fast forward, I used some test tones and figured out what they were using (a boost of side info compared to the mid (or phantom) part. That's about the time Pano started this thread and member CBW posted a link to the LoCo principle (page 8 of this thread).
So I tried that as well, and combined it with what I've learned on this thread.
So I use a variation of that LoCo principle, adjusted to get tonality good/similar between the panned and phantom sounds. This thread was a great help to get me experimenting and realising what did what. I should have some numbers in my thread somewhere. It isn't easy to find the right balance, but if you do: boom! Rock solid image, excellent tonality and holographic on songs that support that. Fun for just about any song. Especially with a Haas kicker brought in.
Your measurement shows a phase offset exactly in the spot where we played with the shuffler. The left/right channels are separated by 45 degree which will alter the cross talk perceived at the sweet spot, but also in the rest of the room.
My mid/side EQ mimics the resulting frequency response of something like that but does not alter the phase. Similar but not the same 🙂.
Cross talk cancelation is another variant to achieve the same. It was easiest to get great results with mid/side EQ. Those results hold up over a larger area without sounding weird.
Thanks for the link to the LoCo pages. I don't fully understand what he's doing, but will study it some more.
When I started this thread 5 years ago my intention was to fix the tonal balance difference between phantom center and hard panned left and right. It's a problem you notice with good speakers in a well treated listening room. A simple artifact of the stereo triangle. It's nice that things have moved on from that to also help with center image stability.
When I started this thread 5 years ago my intention was to fix the tonal balance difference between phantom center and hard panned left and right. It's a problem you notice with good speakers in a well treated listening room. A simple artifact of the stereo triangle. It's nice that things have moved on from that to also help with center image stability.
If I understand this, rock solid might not be the best descriptive term, a person could expect rock solid either way. Rather would pin point be the term to use?
Pin point does not work as a descriptor for me because the phantom image should not be a point. It should be an area.
Rock solid in this case would mean: no wandering/shifting of image positions related to the frequency spectrum used within such a phantom soundscape.
Not sure if I'd describe that as pin point. As Pano said, we're talking about an entire stage and the positions within that stage. I'm certainly looking for something 'life like' within my constraints.
Not sure if I'd describe that as pin point. As Pano said, we're talking about an entire stage and the positions within that stage. I'm certainly looking for something 'life like' within my constraints.
I recall the manufacturer of some common grounding device said one of its its effects was "sharpening of edges" of the sound image locations within the stage area.
Ok, I have no complaints here. I can concentrate on an individual subject within the image, and it's subtleties, to the exclusion of others.
Would you say this frequency dependence produces a steady image with a relative blur, ie a widening of otherwise steady image subjects or would you say the frequency dependence causes variation of position with instantaneous content?
I'll try and describe what this tweak did for my perception, bare with me...
I've used FIR filtering to get my left/right balance in check. Once I did that, the imaging was pretty much focussed, position wise. But it left me with an imbalance tonally. I have to add: this imbalance was after I put up absorption panels at first reflection points.
The panned sounds were perceived as being more bright, the phantom sounds were pin point as one could call it, but 'thin' and somewhat flat (no depth). No body to single entities within the stage. I had a hard time finding a good balance between separate sounds in the stage. It was always a compromise. Prior to the damping panels the sound was pretty forgiving, though a bit wild. I think one of the first things I mentioned after putting up the damping panels and used the FIR correction was: the fir filtering tamed my speakers. A bit too much for my liking.
The first tweak I tried was the JRiver Surround field. Basically this is a boost of the side levels relative to mid within mid/side EQ.
This restored some of the balance differences for the side panned sounds, but more importantly it gave the body back to phantom sounds. Less thin, more realistic sounding. I still had tonal differences but felt it was an improvement overall.
The best part is that it "unlocked" side panned sounds and improved the impression of depth. It also made side panned sounds full bodied (like real persons singing etc.) and no longer as bright. Their relative position changed (more forward, but based on song material) making the stage go beyond borders of the speakers where in some songs the stage was more like 150 degree instead of locked between or imaging in the plane of speaker positions.
This varied with songs though, I did experience imaging like this in my room when I still had more early reflections, however in that situation every song got that same effect. As if the band was 'in the room' with you.
After finding the LoCo tweaks with the "S-curve" in the Sengpielaudio papers, the balance got a little better, but not quite there. The center panned sounds were still lacking in tonal balance, too bright in mid frequencies around 3 KHz and darker above that, which made me adjust the overall room target to reduce that effect.
With this thread in full force, and measured results being added to our scrutiny, measurements made in an excellent room right where the ear position would be, I decided to take it a step further. I adjusted the mid (or center) balance to what Prof Edgar Choeiri's early cross talk compensation simmed like when imported in REW. Making slight 'cuts' at ~3 to 3.7 KHz and more of a 'bump' at 1.8 KHz. 5.5 Khz got a similar (but narrower) bump and another cut, less deep than the one at ~3 KHz, at 7.2 KHz... The sides did not get this tweak, they were left as they were setup in the LoCo S-curve. Below 1 Khz I left the S-curve as is in both mid and sides.
That tweak kept all the positives of the S-curve/Surround field while focussing the phantom parts. One of the first things to notice was an improvement in intelligibility. Finally there was a pleasing tonal balance throughout the stage. But it also set the imaging free to vary between songs. Every entity got it's own defined position with body. Side panned sounds were more real, not as bright (much like with Loco used by itself) and the stage appeared to have depth where it seemed easier to define position even in the front/back plane.
Another addition to this has been the Haas kicker, which, due to room size constraints has been added using ambient speakers. Much of that energy that was there without treatment was back, only more in balance. No longer the "they are here" kind of imaging but it transitioned to "you being there".
While this above may be a hard story to swallow, it took me years to find a balance like that in the real world. To get to that point. And I'll continue to see if I can push the boundaries even further.
After all, I'm looking for (a personal) entertainment value, not a system that I need or depend on to mix music and make decisions. I'm looking for believability. A sense of realness, is that a word? It should be 🙂.
Some might say it's not HiFi, that's about right, this is MiFi, where I'm the judge of the improvement factors. If you read the reviews that some people were nice enough to write up after a visit, I guess I do alright 😉.
Overall I'd recommend to anyone willing to learn how this Stereo game works: Experiment with it. I would not want to go back to a system without this trickery. That's what Stereo does, fool our brain, I just try to make good use of that.
I hope that somewhat answers the question, I may have missed the blur part, no idea what you meant to say with that.
I've used FIR filtering to get my left/right balance in check. Once I did that, the imaging was pretty much focussed, position wise. But it left me with an imbalance tonally. I have to add: this imbalance was after I put up absorption panels at first reflection points.
The panned sounds were perceived as being more bright, the phantom sounds were pin point as one could call it, but 'thin' and somewhat flat (no depth). No body to single entities within the stage. I had a hard time finding a good balance between separate sounds in the stage. It was always a compromise. Prior to the damping panels the sound was pretty forgiving, though a bit wild. I think one of the first things I mentioned after putting up the damping panels and used the FIR correction was: the fir filtering tamed my speakers. A bit too much for my liking.
The first tweak I tried was the JRiver Surround field. Basically this is a boost of the side levels relative to mid within mid/side EQ.
This restored some of the balance differences for the side panned sounds, but more importantly it gave the body back to phantom sounds. Less thin, more realistic sounding. I still had tonal differences but felt it was an improvement overall.
The best part is that it "unlocked" side panned sounds and improved the impression of depth. It also made side panned sounds full bodied (like real persons singing etc.) and no longer as bright. Their relative position changed (more forward, but based on song material) making the stage go beyond borders of the speakers where in some songs the stage was more like 150 degree instead of locked between or imaging in the plane of speaker positions.
This varied with songs though, I did experience imaging like this in my room when I still had more early reflections, however in that situation every song got that same effect. As if the band was 'in the room' with you.
After finding the LoCo tweaks with the "S-curve" in the Sengpielaudio papers, the balance got a little better, but not quite there. The center panned sounds were still lacking in tonal balance, too bright in mid frequencies around 3 KHz and darker above that, which made me adjust the overall room target to reduce that effect.
With this thread in full force, and measured results being added to our scrutiny, measurements made in an excellent room right where the ear position would be, I decided to take it a step further. I adjusted the mid (or center) balance to what Prof Edgar Choeiri's early cross talk compensation simmed like when imported in REW. Making slight 'cuts' at ~3 to 3.7 KHz and more of a 'bump' at 1.8 KHz. 5.5 Khz got a similar (but narrower) bump and another cut, less deep than the one at ~3 KHz, at 7.2 KHz... The sides did not get this tweak, they were left as they were setup in the LoCo S-curve. Below 1 Khz I left the S-curve as is in both mid and sides.
That tweak kept all the positives of the S-curve/Surround field while focussing the phantom parts. One of the first things to notice was an improvement in intelligibility. Finally there was a pleasing tonal balance throughout the stage. But it also set the imaging free to vary between songs. Every entity got it's own defined position with body. Side panned sounds were more real, not as bright (much like with Loco used by itself) and the stage appeared to have depth where it seemed easier to define position even in the front/back plane.
Another addition to this has been the Haas kicker, which, due to room size constraints has been added using ambient speakers. Much of that energy that was there without treatment was back, only more in balance. No longer the "they are here" kind of imaging but it transitioned to "you being there".
While this above may be a hard story to swallow, it took me years to find a balance like that in the real world. To get to that point. And I'll continue to see if I can push the boundaries even further.
After all, I'm looking for (a personal) entertainment value, not a system that I need or depend on to mix music and make decisions. I'm looking for believability. A sense of realness, is that a word? It should be 🙂.
Some might say it's not HiFi, that's about right, this is MiFi, where I'm the judge of the improvement factors. If you read the reviews that some people were nice enough to write up after a visit, I guess I do alright 😉.
Overall I'd recommend to anyone willing to learn how this Stereo game works: Experiment with it. I would not want to go back to a system without this trickery. That's what Stereo does, fool our brain, I just try to make good use of that.
I hope that somewhat answers the question, I may have missed the blur part, no idea what you meant to say with that.
Last edited:
Thanks for this, Wesayso! Really helpful. I will give the mid EQ adjustments a try. The side EQ is the same as the S-curve, is that correct?
Tell us more about the Haas kicker. What is the frequency content, delay, etc.?
Tell us more about the Haas kicker. What is the frequency content, delay, etc.?
Pano, if you look at the "phasey" measurement I posted, it has phase differences in the 1-2kHz range, which is precisely the range where head shadowing occurs and where your original shuffler operated.
Frequency content, I started with L-R and R-L, band passed from ~200 Hz to 3-3.5 KHz (12 db/oct). I ended up mixing in some center later on.
Delay between 15 ms to 20 ms, experimented with shorter delays, but always end up with it around 15 to 17 ms. I delay the center part a little less (about 2 ms) and do not use them as direct devices but let them reflect and scatter from lateral angles. You want them to be decorrelated from the mains. That is the most important for it to work. But you can shape the frequency content/shape
Eventually I added a bit of a reverb tail to them. Not as much so you start to notice them, they are already 12 or more dB down from mains to begin with.
Basically they help hide the real room. After robbing the room from the early reflections I needed something to fill in the missing energy in again.(*)
Reverb is working mostly below 1 KHz, it can bring a sense of envelopment without it being obvious. It enhanced the perception of the front stage. You don't notice it is there specifically. Until you turn it off.
They do double duty as surrounds for Home Theatre (without the reverb).
Inspiration has come from David Griesinger's many papers on how we perceive room sound or spaces.
The reverb I use is Lexicon Random Hall. I did try IR's from actual spaces which was fun as an experiment, but it had a clear signature. The reverb algorithm adapts to the songs and does not show that 'single sauce' on everything you play.
(*) I read a lot of material on the gearslutz forums too, where the general idea usually is to preserve the energy, not simply absorb it.