Inversion or Delay?

[paligap]

How can you tell whether a phase issue is caused by a channel or channels being delayed or being inverted? I'm thinking that if the channel is delayed just the right amount, it could appear to be inverted rather than being delayed. Can you infer anything from the beginning of a song (Is that where a delay would show up?) For example:

Is the LFE delayed there, or just inverted? (Sorry; I don't know how to get a snip with the vertical line.)

 

[HomerJAU]

Inverting will fix it.

@steelydave may comment?

 

[DuncanS]

The LFE should be Low Pass Filtered to limit the HF content, this will delay the signal (which thus may appear to be inverted/out of phase), the Mastering Engineer should compensate for the delay (its a product of filtering, you can't get round the Physics).

 

[paligap]

Another member posted a better snip here:
STEVE HACKETT - GENESIS REVISITED LIVE - SECONDS OUT & MORE (Out in September 2022)

 

[steelydave]

Based on the waveform posted in the other thread, it looks like that mix only needs the LFE inversion.

I don't have extensive enough data to say definitively, but it seems like the LFE delay issue primarily affected some (but not all) mixes done before about 2005. As @DuncanS says, it's a byproduct of the filtering process - my feeling, again unsubstantiated by a ton of evidence, is that sometime around then the plugins or software used to create LFE channels were updated to automatically fix the offset created by the filtering process, ie if it induced at 6ms delay, it shifted the audio automatically by -6ms afterward.

I say this because since then (again, roughly, as I think there are some mixes from the 2nd half of the 00's that still have this issue) generally the problem has almost entirely disappeared - I don't think mastering engineers got wise en masse that it was something that needed fixing, I think advances in technology meant that the problem was no longer a problem at all.

So yeah, my approach generally is if it's a newer mix, I'll check for LFE inversion first, and if it's an old SACD or DVD-A mix, I'd probably look for the offset problem first. The offset one is a little more tricky as you're basically trying to eye-match the ridges in the main channels to the ridges in the LFE channel, so what I usually do is look for a passage that has a lot of bass energy (so I have a lot of visual information to work with) and then verify my results as best I can with a visual spectrum analyser and phase meter like Voxengo SPAN.

 

[bluelightning]

How can you tell whether a phase issue is caused by a channel or channels being delayed or being inverted? I'm thinking that if the channel is delayed just the right amount, it could appear to be inverted rather than being delayed. Can you infer anything from the beginning of a song (Is that where a delay would show up?) For example:
View attachment 85212
Is the LFE delayed there, or just inverted? (Sorry; I don't know how to get a snip with the vertical line.)

From the snippet provided, it is hard to really tell if it is an inversion or delay. For a single tone, a delay of half the period will equate to an inversion. To tell you should look at a patch where there is a burst of activity and more than one frequency.

 

[AYanguas]

...
Is the LFE delayed there, or just inverted? (Sorry; I don't know how to get a snip with the vertical line.)

Yes, when you have the vertical line in Audacity and start the snipping tool, the vertical line dissapears

I follow this procedure:

1) Take a snip from the audio tool (no vertical line).
2) Paste into a new blank slide on PowerPoint, or even in Word.
3) Insert an object (vertical line) in the picture, in the relevant place of a peak waveform. FOR THIS SPECIAL PURPOSE, the verticality of the inserted line has to be Exact!
4) Take again a snip of the edited screen on PowerPoint/Word


Revisiting my previous post :
https://www.quadraphonicquad.com/fo...-more-out-in-september-2022.32625/post-658393
it looks that my inserted vertical line is NOT EXACTLY vertical.

The diagnosis of LFE inverted I think is still valid, because I try to enlarge the wave length picture big enough to easy select peaks and valleys.

 

[MidiMagic]

How can you tell whether a phase issue is caused by a channel or channels being delayed or being inverted? I'm thinking that if the channel is delayed just the right amount, it could appear to be inverted rather than being delayed. Can you infer anything from the beginning of a song (Is that where a delay would show up?)?

The phase canceling delay takes different times at different frequencies.

As the frequency rises, the delay to cause cancellation becomes shorter.

With an inversion, all of the frequencies cancel together.