I must applaud your willingness to participate soongsc!
It's fun to test this yourself. I've used a lot of songs as examples, preferably those that feature the actual artist on background vocals too.
I'm not sure how Pano makes these tests, but compared to a stereo song they probably differ. In a song, like Dr. Geddes said, the vocal part will be processed to get around the side effects experienced in the phantom centre.
A single voice, played in centre and panned to left and right will not have been altered to "fix" that phantom placement.
A real source will sound different too as their presence near walls will change perceived tonality too.
So it's probably going to stay confusing, which is why I played music. In my music I wanted the side panned material to sound like the phantom centre.
Zappa has a lot of vocals coming from both sides and alternated with phantom centre position.
Lorde's Royals has her on backing vocals as well as main. Lot's more songs like that in Pop music. With instruments it's even harder. Especially if there are room queues in the recording, artificially or real.
Listen to songs on headphones to pick out possible candidates. I love the backing vocals on Malia & Boris Black - Claire Cadillac and Steely Dan - Hey Nineteen.
Lots of other examples in recordings from Michael Jackson where he's participating on backing vocals too.
I 'cite' these because of the particular properties of these songs, not as my favourite material. I listen to songs like that as examples. They are not a display of my personal taste in music.
Zappa is a personal favourite though. Usually well recorded.
An instrumental example would be: Rodrigues & Gabriella - Hannuman
Multiple guitar positions which gives you the ability to judge tonality of the ones on the sides compared to the centre panned stuff.
Same goes for the life track of the Eagles - Hotel California of the CD Hell Freezes Over with multiple guitar positions from left to right.
(these two examples differ quite a bit in how they image)
I can name many more that I use, ok one more: Helplessly Hoping by Crosby Stills and Nash, each of them panned in their own position. But not the same voice sadly. So you'd have to know how they sound individually for that one.
It's fun to test this yourself. I've used a lot of songs as examples, preferably those that feature the actual artist on background vocals too.
I'm not sure how Pano makes these tests, but compared to a stereo song they probably differ. In a song, like Dr. Geddes said, the vocal part will be processed to get around the side effects experienced in the phantom centre.
A single voice, played in centre and panned to left and right will not have been altered to "fix" that phantom placement.
A real source will sound different too as their presence near walls will change perceived tonality too.
So it's probably going to stay confusing, which is why I played music. In my music I wanted the side panned material to sound like the phantom centre.
Zappa has a lot of vocals coming from both sides and alternated with phantom centre position.
Lorde's Royals has her on backing vocals as well as main. Lot's more songs like that in Pop music. With instruments it's even harder. Especially if there are room queues in the recording, artificially or real.
Listen to songs on headphones to pick out possible candidates. I love the backing vocals on Malia & Boris Black - Claire Cadillac and Steely Dan - Hey Nineteen.
Lots of other examples in recordings from Michael Jackson where he's participating on backing vocals too.
I 'cite' these because of the particular properties of these songs, not as my favourite material. I listen to songs like that as examples. They are not a display of my personal taste in music.
Zappa is a personal favourite though. Usually well recorded.
An instrumental example would be: Rodrigues & Gabriella - Hannuman
Multiple guitar positions which gives you the ability to judge tonality of the ones on the sides compared to the centre panned stuff.
Same goes for the life track of the Eagles - Hotel California of the CD Hell Freezes Over with multiple guitar positions from left to right.
(these two examples differ quite a bit in how they image)
I can name many more that I use, ok one more: Helplessly Hoping by Crosby Stills and Nash, each of them panned in their own position. But not the same voice sadly. So you'd have to know how they sound individually for that one.
Last edited:
OK, I did casual measurement today. The result clearly shows this phenomenon.
Measured at listening position (Left ear)
Blue - Play Left Only
Yellow - Play Both side (stereo)
Red - Play both side with a small barrier (seat cushion) next to the mic.
Second chart is showing different mic position in dark red, both side playing. (5cm closer to the speaker).
Measured at listening position (Left ear)
Blue - Play Left Only
Yellow - Play Both side (stereo)
Red - Play both side with a small barrier (seat cushion) next to the mic.
Second chart is showing different mic position in dark red, both side playing. (5cm closer to the speaker).
Attachments
I also measured with much larger barrier, and it is very close to the LEFT only chart (blue line) as expected, of course. My current conclusion is still the same as what I have been insisting;
1 The STEREO is fundamentally imperfect, and we're always listening to comb filtered music even in perfect situation.
2 The only solution is a huge barrier separating left and right.
WESAYSO is offering M/S EQ and additional diffused signal, and it would be another solution, but I honestly do not understand his idea yet. I wish he would provide us measured result of his solution.
1 The STEREO is fundamentally imperfect, and we're always listening to comb filtered music even in perfect situation.
2 The only solution is a huge barrier separating left and right.
WESAYSO is offering M/S EQ and additional diffused signal, and it would be another solution, but I honestly do not understand his idea yet. I wish he would provide us measured result of his solution.
Perhaps I don't know what you are measuring or exactly how but your curves - even with an overly smoothed 1/6 - seem very odd to me.
So below is a chart made under crude circumstances showing L, R, and dual-mono, configured to resemble your post. I don't think there's any need for me to label the curves.
As you can see, the dual-mono curve remains neatly above the L and R for the most part. Except at one or two points: if a mic is a few inches closer to one side than the other or if the crossover is a tiny bit off, you'd get that interference but that is quite evanescent, changes inch by inch and note by note, depends on the room and proximity to walls, and has no meaning for stereo listening. And the L and R curves are surprisingly parallel, esp considering this is just my inexpensive vacation cottage loudspeakers somewhat haphazardly placed for spousal approval.
Ben
Attachments
Last edited:
Can you show me 1/6 of your curve to compare with mine? I'm not sure if you can hear this comb filter symptom with your system.
I just want to see your 1/6 oct smoothed out of your measurement curve. Thanks.
Although I have strong preferences for sticking to 1/12, what I posted is smoothed 1/6 to resemble your post. Is that what you'd like?
B.
I see. 12dB curve is attached. The reason why I presented 1/6 is, just because it is clearly show this symptom, which looks close to the simulation on the computer.
I think your measurement is showing some other comb filter caused by the other reasons. WESAYSO and my speaker/room is rather flat when only one speaker is played at a time. I don't do any room correction, so not as good as WESAYSO, though.
Attachments
Thanks for posting that.
How was this measured? A single mic centered between to two speakers?
These dips we see in your measurements are real. I doubt Ben did his best to measure at an ear position like you obviously did so ignore those.
However, in a Stereo recording the vocal material is usually adjusted to account for these dips. At least for the most part. In one of the first pages there's another paper linked which tells you about the S-curve EQ. I suggest to read that and look at the provided info before I try and explain it all. My final mid/side EQ is nowhere near as strong as what they suggest, as mixing and mastering engineers aren't deaf (I hope) and will account for stuff like this in their mix.
But I did notice the effect Pano described in the first few posts. So I started with that S-curve and adjusted it in listening sessions.
The ambient mix I use is described in my thread: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-143.html#post4459831
This started as a simple Haas Kicker, but later on (due to some of the things we discussed here) transformed into a part of my working solution.
If you look at this post from jim1961: http://www.diyaudio.com/forums/multi-way/277519-fixing-stereo-phantom-center-61.html#post4746283
You can see where I got the idea for that. I have a virtual kicker, measurements of that in my thread, but added the ambience (reverb) later.
That's starting here: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-199.html#post4631144
This has gotten another sequel, as I now have the centre part of my ambient signal separated from the regular ambient material. My thread will be filled with these tests and theories, much more elaborate than anything I wrote here.
What Ben showed looks like this at my listening spot:
Oh wow, in the exact sweet spot both signals do sum together. Not a smoothed graph, just a 6 cycle frequency dependent window. I like to see what hit's my ears, not the room stuff, to determine what happens in those critical first milliseconds. Which is why I used the 6 cycle frequency dependent window.
To show that same measurement at 1/12 smoothing for left and right signal without any gating at all:
Posting this one just to show I'm not trying to hide or overly smooth my results. Absorbing the big reflections and EQ based on the ~6 cycle FDW makes the resulting graphs that much closer to that early wave front.
I did not measure at an ear position like you did. As I wanted to get a dummy head for that. Else it will be difficult to really know how head shading influences things.
I've talked to a member here who has such a dummy head, but I could only borrow it (him being from the US). In time I'll probably come up with something, I can actually scan my head in 3D at work. We do not own a big enough 3D printer though. my head doesn't completely fit in the 200 x 200 x 200 mm our printers cover. My head wouldn't be my favourite thing to print either, as it would take quite a while on our simple filament driven printers.
Last edited:
Thanks for that. It clearly shows combing, tho a little different from what we should see with a head in the way. Your barrier lessens the dip, as we would expect, but also shifts them.
Yep, I guess it is because the mic is not exactly positioned. It was a casual test and I placed the mic about 10cm left from the center of the room. The small barrier is causing diffraction, maybe. The Large barrier curve (1mx1m foam) looks much more perfect, but I erased the data by mistake...
And, if you mentioned about the level, 2 speakers are louder than 1, that's why stereo data is louder.
Last edited:
Maybe, in a "by ear" sort of way. Nothing I've read so far in documents about mastering and EQ mention the dull phantom center at all. Not even close. I'll keep reading.
So far I have mostly used voice to demonstrate the effect. I've done that because it's easy to hear on a solo voice. To counter that, I've made a music file with good stereo width and I've panned the vocal across left to right. If your system has the brighter sides, you should hear that in this track as well as the counting tracks.
https://drive.google.com/open?id=0B...l2Z2l4eURpNzlBU0luS25qUzY2blgxRVBfOFZVUjdrbTQ
Look for the Dat Dere files. Ones is normal, the other is shuffled. There are some voice only files, too - male and female.
This is what you will get with amplitude panning and no EQ applied to even out the tonal balance between left, center and right. It is possible that because of reflections or other factors, someone may not have the effect on his system.
I'll keep looking and reading to see if anyone in the mastering world is specifically addressing this. So far, I have not found anyone talking about it.
WESAYSO, I'll reply to your interesting post later, I have to make dinner tonight.
Hahaha, fine, I tried to search for that paper I mentioned, but that link seems down, it was the LoCo link in this post: http://www.diyaudio.com/forums/multi-way/277519-fixing-stereo-phantom-center-8.html#post4416556
I might have saved the relevant data, I'll look for it.
Yes, single mic, where my head would otherwise be.Thanks for posting that.
How was this measured? A single mic centered between to two speakers?
Please let me know if I'm missing some important test feature. I don't want to inadvertently post misleading information.
B.
Don't bother. A small - kid size - soccer ball will do the trick. They can be found at most sporting goods stores. A standard football (soccer) is too wide. A ruby ball in the vertical orientation might also do the trick.
Well based on how tonality is set-up you could experience a dull centre or bright sides
.With everything still setup for regular stereo recordings, 5.1 material sounds dull in the (phantom) centre with my current solution. That does need a different EQ scheme.
I'd agree this "correction" I spoke of isn't top notch all the time. But things like reverb etc. help hide the perceived tonal balance too.
On some songs, a bit of PEQ based on the standard dip pattern we see actually improves intelligibility by quite a margin.
So it's save to say on those songs nothing was done to help counter that phenomenon. It's why I have a Christina Aguilera CD in my test collection.
From the album: Stripped
One example song that sounds normal, in other words something was done to counter the phantom problem: Impossible
This one sounds good with perfectly understandable lyrics.
The next song: Underappreciated - this one isn't fixed i.m.h.o. EQ the dip pattern and you can hear the words she uses way better. You can hear the difference between the 2 songs even on headphones. Just a few simple cuts and boosts to counter the cross talk effect helps this second track.
That album has a surprising amount of neat imaging tricks and other effects. I did not expect to find that. I don't even remember what made me listen to it.
A lot of these songs work as a good stereo demo though. Very different from track to track.
Over time my EQ on the phantom centre got quite flat throughout. But this might be because the Haas Kicker is helping out too?