@markhadman: While it might be more technically correct to refer to polarity instead of phase in this context, the ‘phi’ symbol (which is often used for phase) is generally accepted (rightly or wrongly) within the industry as a standard symbol on pro consoles. Having worked for many years for one of the top pro-audio console manufacturers, we always used the ‘phi’ symbol, and everyone just knew what it meant.
@macinnisrr: Phase seems to be one of the most commonly misunderstood concepts - this is probably not the best place to give a detailed explanation, but briefly, what you describe e.g. time shifting a region, is better termed a delay. In simple terms, for a sine wave, a delay can equate to a phase shift of n degrees, but consider what happens if the same delay is applied to a sine wave of a different frequency. Then you will have a different number of degrees of phase shift (you actually find that the number of degrees of phase shift caused by a fixed delay is linear with respect to the frequency). When you invert the signal you are not shifting it in time at all, you are just multiplying all the sample values by -1
the other issue with terminology for this is that the same terms are used to cover two very different situations. the more common one, i think, is when using multiple mics on the same signal source, with spacing differences between the mics causing phase differences in the signals they pick up. the classic example is a drumset where a mic “in front” of a drum skin is 180 degrees out of phase with respect to one “behind” a drum skin. it just so happens that 180 degrees out of phase can be corrected by signal inversion, but that’s the only phase relationship that can be corrected in this polarity-inverting trivial fashion.
the less common one, i suspect mostly because the majority of people don’t do their own wiring, is where you actually have the polarity of some analog electrical signal path wrong. the trivial case would be a mic cable that is wired “backwards”.
One of the other uses for a signal inversion is in the creation of a ‘mix minus’ - where you can effectively use the signal inversion to subtract it from a mix by cancellation.
While you are correct in that it is termed as a delay, what that delay causes is phase cancellation based on the frequency, and thus why phase shift gets commonly used as I am sure you know(More for other’s benefit)
In my experience Mix-Minus mixes aren’t necessarily made via cancellation like you described, but by creating a completely new mix instead. I deal with them primarily in the installation and programming of automixers and installed systems, but the same basic concept would hold true in most cases I believe, though would love to be corrected on that if you know of examples where it actually is applied via cancellation.
I do agree the ‘phi’ symbol is accurate to use int his case at it is the commonly accepted symbol on pretty much any console(I actually thought until you correct me there it was theta, oops;) However the terminology should be corrected in all documentation to describe it.
Technically even your example is not a good example as it would only apply if both microphones are identical in frequency response, spaced perfectly, and if the signal coming into them was identical. That last part won’t happen period, the frequencies picked up by the bottom mic are different from the top mic, and very rarely are the spaced perfectly identically. So without the signals being identical it can’t truly be described in terms of phase without breaking it down into frequencies.
Actually I would suspect the far more common occurrence IS when the polarity is reversed in wiring when dealing with audio. This doesn’t apply as much to the recording environment, but still applies to there commonly enough. In the live and installation environment it is FAR more common to be a wiring problem. But it applies in other ways as well, including with speaker coverage design, live reinforcement delay timings, etc.
At any rate the short of this is that the main point still applies, Polarity Inversion is the correct descriptor for what Ardour does.
@seablade: mix minus mixes can be created in a number of ways, in many cases it is preferable to construct an entirely separate mix rather than rely on cancellation to exclude a signal. As I’m sure you know, typically mix minus mixes might be used more in live or broadcast setups (though it might be applicable, for example if recording a podcast with ardour, with several remote contributors, each of whom require their own mix minus ‘clean feed’). On some pro format consoles the mix minus is derived by inverting and routing to a mix minus bus. There’s some info about this method on a Euphonix console here:
That is interesting to read their implementation, my usage of mix-minus has always been to build the mix from scratch which in general I would find a better solution for several reasons most likely, but good to know other options exist on certain consoles.
@seablade: Building a mix from scratch is preferable in many cases - as usual with these things there are several approaches to get the same result and neither one is necessarilly technically superior in all cases.
@Paul: While your original example is a good illustration of where / when a phase inversion can fix a problem, typically in multi-microphone setups, the problems are caused by the slight differences in time taken for the sound (or reflections of it) to reach the different microphones. As mentioned in my original post, this delay between the two signals causes a phase shift which is related to the frequency of the signal. When the two signals are re-combined in the mixer, you can get a comb filter effect as the signals sum and cancel at different frequencies. This is similar to the deliberate effect created by a flanger (though not modulated). A typical example might be if you recorded an acoustic guitar and vocal, each might sound fine, but when you mix them together the cross-talk from the vocal and guitar mics adding with their respective original signals might cause problems. Inverting one of the signal sources can help reduce the audible effects of this in some cases, other solutions are to time shift the track in a DAW, or use one of the analogue phase adjustment boxes (the IBP - In Between Phase box springs to mind) or even use the emulation of such analogue phase alignment boxes that I’ve built into my gate plugin
@linuxdsp: right, i was referring solely to the simple case of actual 180 degree phase differences, which as you and seablade have noted, are quite rare. i was not trying to suggest that most typical recording configurations lead to such trivial differences - most of them are tricky to fix as you noted.
This is a fabulous announcement and overview. Thank you very much!
I hope that in future there will be more news updates like this, showing progress with development.
After a year of donating $10 each month, I didn’t renew my subscription because Ardour 3 hadn’t been released, and lack of communication regarding progress and its status left me wondering what I was donating for.
However, this announcement is well written, detailed, and exciting, and I can’t wait to try 3.0 and encourage as many other people as I can to do the same!
I understand only samtuke wrote:
After a year of donating $10 each month, I didn’t renew my subscription because Ardour 3 hadn’t been released, and lack of communication regarding progress and its status left me wondering what I was donating for.
For donation I want to see a roadmap and the possibility to download alpha , beta and finaly release candidate.
But I will wait no problem it is worth to me.
