argh! it's polarity not phase!

[timlloyd]

The Polarity invert button is labelled Phase invert...................it should not be!

[EvilDragon]

Err, no? "Phase inverting" is a normal all-around term in DAWs, no need to change it. It does what it says on the label, end of story. Also, "phase" is more related to audio waveforms than "polarity" in general. Nothing wrong here.

[timlloyd]

I disagree but whatever That's why it's in "nitpicks".

[bogo]

Originally Posted by timlloyd

The Polarity invert button is labelled Phase invert...................it should not be!

I agree. +1

[timlloyd]

Here you go ED

[drybij]

Doesn't 'polarity' imply 'polar' as in magnets or in DC current flow? Seems to me phase makes more sense when talking about waves or AC current.

[timlloyd]

It depends on what exactly you're referring to when talking about audio waves.

Audio signals ARE bi-polar. If a speaker can reproduce it audibly, it contains positive and negative voltages and is an AC signal.

Phase shift refers to time and is dependent on wavelength, polarity does not and is not.

They both make sense when talking about audio waves and AC signals, but they're not the same and they refer to different things.

They are only colloquially interchangeable at best.

[Mr. Green]

It seems to me that it all really depends on your perspective of what is going on when you press the "phase" button.

From the point of view of the track as an isolated thing, you are indeed altering its' polarity. However, when viewed as a single part of the whole stream of audio, you are in fact altering it's phase in relation to the other tracks.

lol - whuuuuuutever, right? ;-)

[DuraMorte]

Originally Posted by Mr. Green

From the point of view of the track as an isolated thing, you are indeed altering its' polarity. However, when viewed as a single part of the whole stream of audio, you are in fact altering it's phase in relation to the other tracks.

My opinion as well; leave it alone, it serves it's purpose just fine.

[beatbybit]

Leave the Phase invert button alone!!

[timlloyd]

Originally Posted by Mr. Green

From the point of view of the track as an isolated thing, you are indeed altering its' polarity. However, when viewed as a single part of the whole stream of audio, you are in fact altering it's phase in relation to the other tracks.)

No, you're still altering its polarity. There is no phase shift going on whatsoever. It may reduce/increase phase cancellation, but in and of itself it is NOT a phase shift.

Phase is relative time-difference - press the "phase invert" button and what you're doing IS NOT shifting that audio backward/forward in time. ALL you are doing is multiplying every sample value by -1 - that is inverting the polarity of the signal.

The term "phase invert" requires additional information to make sense. You're shifting all frequencies forward/backward in time by some amount. Are all frequencies being shifted by the same amount of time? Which frequency is this amount derived from? The detected fundamental frequency of the audio stream? The lowest frequency? The highest frequency?

The time that each frequency needs to be shifted by in order to be 180 degrees "out of phase" relative to before processing is directly correlated to the wavelength of that frequency.

If you gave every frequency a 180 degree phase delay, the audio would sound completely different because every frequency would be delayed by a different amount of time.

Only when talking about two signals of the exact same frequency does "phase invert" make any sense.

If you delay an audio stream by the same amount for every frequency, relative to another audio stream, you would be creating a time-difference and would call it a delay. It's only "phase difference" if the two audio streams are of equal frequency (waveform can differ - frequency must be the same).

If I have a guitar track and a drum track and I delay the guitar slightly, it *is not a phase-difference*. It's simply a delay!

Does anyone understand what I'm going on about?!

*sigh*

[timlloyd]

Daammmmmn guys, I'm not suggesting anyone should change the functionality of it!!

It's simply labelled incorrectly............regardless of the fact that it's an oft-sprouted misnomer...........it's misleading and inaccurate...........which is why it's in "nitpicks" not "bugs".

jeeeeeeeeeeeesus

[zappsunzorn]

http://www.rane.com/par-p.html

polarity A signal's electromechanical potential with respect to a reference potential. For example, if a loudspeaker cone moves forward when a positive voltage is applied between its red and black terminals, then it is said to have a positive polarity. A microphone has positive polarity if a positive pressure on its diaphragm results in a positive output voltage. [Usage Note: polarity vs. phase shift: polarity refers to a signal's reference NOT to its phase shift. Being 180° out-of-phase and having inverse polarity are DIFFERENT things. We wrongly say something is out-of-phase when we mean it is inverted. One takes time; the other does not.

[drybij]

got it - thanks for edumacating me!

[Dannii]

Originally Posted by zappsunzorn

http://www.rane.com/par-p.html

This is the core of the matter. Phase = time, polarity <> time. Most physical mixing consoles have this wrongly labelled too.

Perhaps I might go and redesign the polarity buttons in my REAPER themes

[musicbynumbers]

I too agree that it should be labelled "invert polarity" or something along that line. Not "phase", let's set a standard here! lol

Whilst the devs are at it perhaps for reaper 4 we could have a phase alignment dial (one that you can turn off to save dsp)

[Fabian]

Tim is right here, of course, polarity and phase are different things. The "invert phase" button does not invert the phase, but the polarity. With plain sinus signals, polarity inversion is the same as 180 degrees phase shift (as shown by the pdf Tim gave earlier). With more complex signals, the relation between polarity and phase is much harder, if at all possible, to define.

That said, I'm not sure "phase inversion" really has a meaningful definition, not even for plain sinus, so using that term to mean "polarity inversion" could be acceptable...

[Dannii]

Originally Posted by musicbynumbers

I too agree that it should be labelled "invert polarity" or something along that line. Not "phase", let's set a standard here! lol

-1.... oops, wrong polarity!! I meant +1..... *red wire to pin 2, white wire to pin 3*

Originally Posted by musicbynumbers

Whilst the devs are at it perhaps for reaper 4 we could have a phase alignment dial (one that you can turn off to save dsp)

AKA micro delay with positive and negative delay amounts..... just to confuse the issue!!!

[Ollie]

[Tries to imagine a Fender Twin with a Tremolo labeled "Tremolo"]

[JHughes]

I'm glad I'm not the only person bothered by this. Polarity is not phase, no matter how many high end pieces of gear get it wrong.

That's why it's called audio "engineering", because it really does help you do your job better if you actually understand this stuff.

With plain sinus signals, polarity inversion is the same as 180 degrees phase shift

Not at the beginning and end of the waveform!

Please change the labeling! (Dannii, there's no reason for you to wait on Cockos to wake up.)

[fuseburn]

Originally Posted by Ollie

[Tries to imagine a Fender Twin with a Tremolo labeled "Tremolo"]

hehehe good one !

I'm with the OP. Still, I'd think that a short sidenote in the manual would suffice.

[timlloyd]

Originally Posted by Ollie

[Tries to imagine a Fender Twin with a Tremolo labeled "Tremolo"]

Haha true.

but still..............it's silly

[liteon]

it is accurate to display "polarity invert" instead of a "phase invert" for a discrete signal, so yes theoretically correct, but i guess most digital mixers and daws have it for sentimental reasons and also to avoid semantics and correspond to customs.

but how do you achieve polarity inversion in the real world (analog) and what is the importance of this study?

first of all the "phase invert" label on your analog mixer (being neve or ssl for example) is technically accurate. polarity inversion in this case is time dependent and in fact due to a phase shift.

the article posted above does differentiate the terminology quite well, while giving examples in acoustics, but fails to explain that polarity inversion is clearly mathematical (theoretical) therm and yes dsp does exactly that. while it works in dsp this does not mean the same would work for an analog circuit as it is impossible to have instantaneous polarity inversion for continuous signals with a "static multiplier" - or at least i do not know of such technology.

most likely, your analog mixer's "phase invert" control being based on on semi-conductor or vacuum tube technology, will have to obey some very simple laws of physics as far as signal manipulation goes. there is evidently going to be some phase shifting occurring at such a circuit block.


--

[JHughes]

Originally Posted by liteon

but how do you achieve polarity inversion in the real world (analog) and what is the importance of this study?

For an XLR swap pins 2 and 3, for an unbalanced signal swap tip and sleeve.

Originally Posted by liteon

first of all the "phase invert" label on your analog mixer (being neve or ssl for example) is technically accurate. instantaneous polarity inversion in this case is time dependent and in fact due to a phase shift.

I'm confused now. Polarity inversion on an analog console has nothing to do with time (it's instantaneous) or phase (phase is unchanged, that is, the start and end points of the waveform remains exactly the same).

[ubergod]

So a phase shift of, oh, 35 degrees delays the audio?

If I used one of these things:

http://www.littlelabs.com/ibp.html

...and turned it to 180 degrees, how much would I have delay one of the signals so it cancels out completely?

[ubergod]

Originally Posted by musicbynumbers

Whilst the devs are at it perhaps for reaper 4 we could have a phase alignment dial (one that you can turn off to save dsp)

There is a JS plugin included that does that, no?

Utility/phase_adjust

If this one is set to 180 degrees, it definitely cancels out complete with a copy without the "phase adjust"

so... what is it folks?

[timlloyd]

Originally Posted by liteon

first of all the "phase invert" label on your analog mixer (being neve or ssl for example) is technically accurate. polarity inversion in this case is time dependent and in fact due to a phase shift.

There is no phase/time shift when you hit a polarity invert switch on an analogue console. Like JHughes said, you just swap + and -. If it was due to a phase shift, which frequency present in the input signal is the time-difference calculated from? It wouldn't matter which, because only that single frequency would be 180 degrees out of phase with itself, the others would not.

And even for that single frequency, it wouldn't be the same as a polarity inversion, because a phase shift took place, meaning that the frequency was delayed in time. There is no time-delay in a polarity inversion.

Originally Posted by ubergod

So a phase shift of, oh, 35 degrees delays the audio?

If I used one of these things:

http://www.littlelabs.com/ibp.html

...and turned it to 180 degrees, how much would I have delay one of the signals so it cancels out completely?

How much time difference is required to produce a 180 degree phase shift is dependent on frequency.

Yes, phase shift does delay the audio.

This happens in EQ as well; that's how they work (excluding FFT-linear-phase).

You delay a signal by a certain amount and it causes cancellations/reinforcements at certain frequencies.

Phase is time.

Polarity is not time.

Polarity inversion does not affect time.

The Little Labs IPB affects phase. It delays signals. It does this using two symmetrical all-pass filters. It is not polarity.

There is no such thing as a "phase invert button". Yes it's semantic pedantry to some, but to others it's a glaring wrong-un of confused terminology

[tls11823]

How about if they call it "phasarity"? Then everybody will be happy.

[ubergod]

Originally Posted by timlloyd

How much time difference is required to produce a 180 degree phase shift is dependent on frequency.

Yes, phase shift does delay the audio.

This happens in EQ as well; that's how they work (excluding FFT-linear-phase).

You delay a signal by a certain amount and it causes cancellations/reinforcements at certain frequencies.

Phase is time.

Polarity is not time.

Polarity inversion does not affect time.

The Little Labs IPB affects phase. It delays signals. It does this using two symmetrical all-pass filters. It is not polarity.

There is no such thing as a "phase invert button". Yes it's semantic pedantry to some, but to others it's a glaring wrong-un of confused terminology

I understand all that, was just asking if you could explain why/how a phase shift of 180 degrees seems to result in the same thing as a polarity inversion.

[Lokasenna]

Maybe it's the small amount of calculus I took, but "phase" still makes perfect sense to me even if it's technically inaccurate.

If you consider it to be shifting the instantaneous value of the tiniest chunks by 180', thus inverting each instant rather than shifting the whole signal, does that not clear up the discrepancy?

I know it's not actually what's going on, but it seems reasonable to me.

At any rate, I defy you all to come up with reasons not to just label the button "Invert" and be done with it.

[beatbybit]

i think its time for a FR:

user def. tooltip on that button.

[zappsunzorn]

Originally Posted by beatbybit

i think its time for a FR:

user def. tooltip on that button.

But the default should be polarity

[beatbybit]

Originally Posted by zappsunzorn

But the default should be polarity

or "whatever this button called" ?

[JHughes]

Originally Posted by Lokasenna

If you consider it to be shifting the instantaneous value of the tiniest chunks by 180', thus inverting each instant rather than shifting the whole signal, does that not clear up the discrepancy?

No. A shift is not an invert, and if you shift the tiniest chunk by 180, that only applies to one specific frequency out of 20,000+ integers. So which specific frequency do you choose to "invert" at the expense of all others?

[Dstruct]

Same topic: http://forum.cockos.com/showthread.php?t=24084

[ubergod]

lol, I made this gif, can't believe it's still up on shack:

[img]http://img396.**************/img396/639/phasezk4.gif[/img]

[beatbybit]

mine lost it seems..

anyways, can you see the "bug" in that plugin GUI? its shifting.. phase? as you rotate..

[timlloyd]

No. No compromises. Schwa's compromise in that other thread is still incorrect.

argh

Originally Posted by ubergod

was just asking if you could explain why/how a phase shift of 180 degrees seems to result in the same thing as a polarity inversion.

It only results in the same thing when you have two identical periodic signals, and discount the fact that with a 180 degree phase shift, one of the frequencies will have been delayed/advanced in time by half its wavelength. You would not notice this with two identical periodic signals, but it would be very obvious with a practical example.

The fact is, when you use your polarity invert button to "phase align" a multi-mic recording, you're not 'phase aligning" at all. All you're doing is switching the positive and negative portions of a signal so that it being out of phase with another causes less destructive interference when they sum.

Bloody nitpicks sub-forum, getting me all hot-and-bothered about terminology. I should throw away my laptop and pick up my guitar

[liteon]

Originally Posted by JHughes

For an XLR swap pins 2 and 3, for an unbalanced signal swap tip and sleeve.

yes, i guess in this case a DPDT switch can be used and it does the same as swapping poles mechanically. its still not as instantantanious as the DSP "-1" multiplier (there was some sort of small signal decay if i recall correctly).

Originally Posted by timlloyd

There is no phase/time shift when you hit a polarity invert switch on an analogue console. Like JHughes said, you just swap + and -. If it was due to a phase shift, which frequency present in the input signal is the time-difference calculated from? It wouldn't matter which, because only that single frequency would be 180 degrees out of phase with itself, the others would not.

the term you are looking for is "absolute phase" of a system.
all frequencies can be 180 degrees out of phase in an ac signal with a npn transistor - due to change in base voltage, which results in 180 degrees phase change in collector voltage or the more "composite" closed-cycle op amp - due to its feedback loop / inverting buffer.

here is an example of a "phase invert" switch (with some audio applications i believe):
http://sound.westhost.com/project107.htm

---

both of you have valid points and i agree with them, but the subject itself is quite complicated.
i personally wouldn't mind if the label of a daw or an analog console is inaccurate for this specifically.

---

[ubergod]

Originally Posted by timlloyd

It only results in the same thing when you have two identical periodic signals, and discount the fact that with a 180 degree phase shift, one of the frequencies will have been delayed/advanced in time by half its wavelength.

So why does the 180 degree phase shift completely null out with the non-phase shifted one, and also completely null with the one that is "reverse polarity"? I'm not arguing with you about the terminology, just trying to understand this stuff better. Also, no worries if you can't get me to understand it, it may just be way over my head. Thanks for your help so far