Phasing problem (Mastering)

[mischmerz]

Hello -

I was asked to listen to a few transferred stereo vinyl recordings. I fed it through my tool chain and noticed that my "Jellyfish" (Goniometer) showed large parts of the recording to be in the "red" - indicating canceling (stereo) phasing. I know that the guy didn't do anything fancy during transfer (Vinyl -> RIAA -> Tascam Digital Recorder) and that the vinyl records in question are high-end commercial manufactured. I checked a few of my own transfers and .. well ... I never even see a touch of "red" on the display.

Is the "red" on the "jellyfish" display considered to be normal for some vinyl records or is it always and indication of a problem?

Thanks for your insight.

Michaela

jelly.jpg

[Xertz]

My guess would be bad stereo separation between the two drivers on the cutting head itself.

[mischmerz]

So - you would suggest that it was badly cut? Not a reproduction issue?

m.

[gold]

There is nothing wrong with uncorrelated material. That is what stereo is. As long as there isn't so much vertical modulation that the record will skip it's no problem. The meter is not frequency sensitive so that reading could represent slightly uncorrelated audio anywhere in the frequency spectrum. Uncorrelated low end can be a problem at high levels or very low frequencies. I've cut stuff that would peg that meter at -1. All the way red. But it is all high frequency audio so no problem to track.

[mischmerz]

@gold - As far as I understand that - doesn't it indicate that sound waves are canceling each other out? In other words - one would hear *less* ?

[gold]

If you have two identical signals that are 180º out of phase it's called reverse polarity and the signals will cancel. The meter would read all the way -1 and never move. If you have two signals that are identical and you sum them there will be a 6dB level increase.

If you have two signals that are not identical, for example a 100Hz sine wave and a 200Hz sine wave, that are reverse polarity they will not cancel. Music is almost always a complex waveform. The phase relationship and subsequent cancellation and summing is what makes stereo work. That is what stereo is. Without that it would be mono.

The -1 to +1 meter is called a correlation meter. It is showing what is common or not common to both channels. If the meter reads +1 it means both channels are identical (mono). Anything that isn't +1 means there is a difference between the left channel and the right channel. If the signals are identical but reverse polarity the meter will read -1. Anywhere in-between +1 and -1 means some portion of the signal is different between left and right.

@gold - As far as I understand that - doesn't it indicate that sound waves are canceling each other out? In other words - one would hear *less* ?

Less than what? If that's the way the music is that's the way the music is.

[Xertz]

mischmerz - Yeah, well that's what I originally thought, but gold is probably right about what he said in last post. gold - I don't think you understand what i'm saying. If one pushrod is just a tiny bit slightly smaller than the other or something similar than I think that could cause phase issues like with what mischmerz is describing.

[gold]

mischmerz - Yeah, well that's what I originally thought, but gold is probably right about what he said in last post. gold - I don't think you understand what i'm saying. If one pushrod is just a tiny bit slightly smaller than the other or something similar than I think that could cause phase issues like with what mischmerz is describing.

Is there an issue besides not liking the way the picture looks?

[Xertz]

Ah. I should of read your last post more carefully. Now I understand what your are talking about. Sorry for the confusion.

[mischmerz]

Thank you all for the input. I spoke to another audio-engineer friend and he said that a red (negative) value in an otherwise good cut or transfer first and foremost indicates what would happen if you were to sum up the channels to mono. Everything from 0 to +1 can be combined into a good mono signal, while red would indicate losses. He pretty much confirmed @gold's view but added that _in his experience_ a very "red" correlation meter often indicates an artificially widened digital stereo image.

So, thanks again for you input.

Michaela

[diamone]

As long as there isn't so much vertical modulation that the record will skip it's no problem.

A very "red" correlation meter often indicates an artificially widened digital stereo image.

Or matrix quadraphonic (SQ, QS, EV-4, Dynaquad...) or a stereo test record from 1956-7.

In either case a lot of younger mastering engineers are not aware that a lot of these records have a vertical deviation equal to and in a couple cases greater than the lateral deviation.

And then there's been a considerable number of cases where the original 3-track half inch master had only ONE of the three channels out of phase. I have one copy of a 1957 Capitol demo tape `The Stars in Stereo' where the center track and the left track is out of phase and another copy where the phase is correct across all three.

On the out-of-phase copy if you flip to mono the band is intact but you lose half the choir. I also found that some UK pressings of the same LP retain the phase error. In those cases, the vertical displacement looks like the Himalayan Mountains - even compared to a matrix quadraphonic LP of the 70s - demonstration versions of which have an even greater vertical displacement than the commercial releases.

The other place I have seen large vertical displacement on an LP is on the CBS and Denon test discs (and a few release pressings) for CX and dbx noise reduction respectively.

So they are out there.

I would check around and see if it was originally meant to be that way or whether or not the original master tape or session work parts can be located to check vs going with a needle-drop in the first place.

[Greg Reierson]

What does it sound like? Lots and lots of music has wide, uncorrelated stereo content. That's what makes it sound wide. Not a flaw. A feature. That's how stereo works.

Let's assume it's correct until it sounds wrong, not when a meter looks funny.

[Greg Reierson]

A very "red" correlation meter often indicates an artificially widened digital stereo image.

Or matrix quadraphonic (SQ, QS, EV-4, Dynaquad...) or a stereo test record from 1956-7.

Or a classical recording.

[mischmerz]

@Greg - can you please elaborate a bit? I'd like to understand that on a more detailed level. I assume it may have something to do with the positioning and spacing of the microphones (classical concert recording) ? Or is the "effect" created in the mastering stage?

Michaela

[gold]

If you picture a concert hall and the sound of an orchestra a +1 reading on a correlation meter would sound like all the sound is coming from the conductors podium. A single or mono point source. The contra basses are house right and first violins are house left. That’s stereo.

[Greg Reierson]

The vast majority of studio recording is done with mono mic techniques so the stereo most people are familiar with is panpot stereo. The stereo effect is due to differences in level between the left and right speaker and will usually look very narrow on a correlation meter. Most classical recordings use various stereo mic techniques that create stereo based on phase, arrival time and frequency differences between the mics. They will look much less correlated on a meter.

Imagine making two recordings of the same orchestra with two sets of mics. The first set is a pair of spaced omnis. The second is an XY cardioid pair. They will both capture a stereo recording but the sound will be different. The recording made with spaced omnis will be wide and show very little correlation on a meter. The stereo effect is due to timing differences between the mics. The recording made with the XY pair will be more focused towards the center and show more correlation on a meter. Since the mics are coincident (in the same physical space) there is no difference in arrival time so the stereo effect is due to frequency response differences of the two mics.

The only time any of this matters in cutting is when, as Paul said above, excessive uncorrelated LF energy may cause tracking issues. Any mono compatibility issues beyond that are up to the producer, not the cutting engineer.

[mischmerz]

Thank you all. It think I understand it better now. I was confused because 90% of my work is with vintage mono and stereo recordings and I so far I didn't notice any "red" on the correlation meter, not even with older 50s and 60s classical recordings. So I was somewhat surprised seeing that much "out of phase" with this recent classical recording. It sounded very good of course, spread very wide, but the center was little lost. I kind of prefer a sound "image" that spreads out from a well defined center to the left and right, but that seems to be a matter of taste

Michaela