It depends. Half a millisecond is, I believe, around 7" worth.

However, interferometry teaches us that small misalignment between drivers and crossover implementations would be completely audible even at a few millimeters.

So, matching your tweeter to midrange to woofer electroacoustically is critical.

Despite many studies we have never seen consistent acceptance or demand for speakers with perfect phase/delay response.

What is my point to all of this?  That delay mismatches are complicated.

I've been saying for years speakers must be precisely equidistant and symmetrical. The researchers could have saved an awful lot of time and money simply asking me. Seven inches, ha! Even a fraction of an inch has an audible effect on image focus.  

Don't need no anechoic chamber. Don't need no measurements at all. Simply set your speakers up with tape measure and framing square to be perfectly symmetrical and equidistant. Listen. Then move one an inch or so. Listen again. 

Once again the reality of listening proves superior to measurebating.
millercarbon, you need to make up your mind. In regard to subwoofers you have stated that "time does not matter." 

When it comes to crossovers it matters a lot. Group delays can be caused by positional factors and electronic factors. You can adjust any set of group delays to be exactly the same. This can be done by moving the slow drivers towards the listener or digitally delaying the early drivers to match the late ones. 1 ms = 1 foot (at sea level). I can usually tell something is wrong by 0.2 ms at low frequencies but have never blinded myself to this so it could be longer. I can tell the difference in bass impact but I am the one adjusting the delay. Of course I set them all to zero.
The vast majority of audiophiles do not have this capability which IMHO is a shame. It is vital for proper subwoofer management and imaging in general. It is the difference of hearing one vs two voices. 
I waited a couple weeks to get my Herron, almost two months to get my Moabs, and now waiting on DHL to deliver my Origin Live Sovereign turntable. Oldhvymec waited even longer on his Decware ZP-3. And that is nothing, after saving for more than a year for a Raven Reflection when finally ready to order was told it will be more than a year! So yes someone somewhere definitely needs to pay more attention to delays in sound!
What mijostyn said about group delays and time delays is correct. It was discovered in the 1930's, by MGM studio.

Below article from The Klipsch Audio Community:

This entire thing goes back to the Eleanor Powell "double tap" incident at MGM studios in the early 1930s. When monitoring playback of Powell tapdancing an echo was noticed on the taps. John Hilliard, subsequently the greatest of all horn designers, was a young sound engineer at MGM and traced the problem to the 8 foot difference in the path lengths of the horns in the 2-way Western Electric monitor speaker.

Hilliard did experiments that showed that the effect was time and frequency dependent and that a delay of less than 3 milliseconds (about 3 feet) was inaudible using crossovers between 350 to 800hz.

A result of this incident and WE's refusal to market their improved Fletcher loudspeaker was the decision of MGM sound honcho Douglas Shearer for MGM to design their own improved system which would among other things minimize time delay. Hilliard was put in charge of the project which then developed the famous Shearer Horn. An entire galaxy of talent worked on this system including Hilliard, James Lansing, Bob Stephens and even RCA's Harry Olson.

Note that though the folded bass horn of the Shearer cut time delay compared to the WE snail horns it didn't eliminate it totally but Hilliard's later Altec Voice of the Theatre did. So it seems that even though Hilliard thought some delay was inaudible the existence of any at all still nagged at him.

Which brings us back to the present day. Is the time offset of the big Ks a problem? Theoretically the distance falls under the distance Hilliard thinks was undetectable. Might it still be a problem? Perhaps. But keep in mind that all loudspeaker designs make compromises, given a choice, as PWK was with his design, between lower distortion and time delay which would you choose? I'd take the lower distortion of the horn system.

Photo of Shearer Horn from Lansing Heritage site
Isn't this thread really talking about a speaker design issue that is inherent in most if not all speakers: namely time coherence.  A couple of manufacturers address this issue by careful placement of the drivers and using first order crossovers. For example, .... Thiel, Vandersteen and Green Mountain audio.  

The late, great Roy Johnson of Green Mountain audio wrote some pretty impressive white papers on the subject.  Not sure if Roy's white papers are still available.

And DEQX sells a digital device that can correct for time delays in speakers.  I used to own a DEQX. The unit was inserted between my preamp and amp.  Although the DEQX added quite a bit to the SQ, I found, on balance, that there was a net subtraction because the DEQX added a lot of processing and manipulation to the signal. 

If there were more brick and mortar stores around that carried a lot more speakers brands, I would love to spend some time educating my ears to learn just how much better time coherent speakers sound as compared to those which are not time coherent.  John Rutan at Audio Connection, located in Verona, NJ, is a Vandy dealer and stalwart.  Perhaps John would let me burn some time one day in his shop.  

There is also Enhanced Bass Alignment (EBA) developed by Meridian to address group delay.
@bifwynne buy a pair of full range single driver speakers and be amazed(I know I was) at perfect time coherence. My recommendation for something not to expensive is Audio Nirvana. 
That’s a remarkable measurement, didn’t know we didn’t know it.

3 things seem related

1. time alignment of frequencies, highs traveling a speck faster than lows. Some designers slope the face of their enclosures for this purpose. You can simply tilt/lean your flat faced speakers back to ’simply’ achieve that. It often directs the tweeters dispersion to your seated ear height, another benefit. Another benefit is the altered angle of reflections off room surfaces

2. cd players. most use a single DAC to process left/then right; left/then right ... I forget the speck of time delay this introduces, to EVERY l/r bit of conversion. To avoid this speck of time delay. better players provide separate L & R i.e. my first player, Onkyo had a matched pair of burr brown processors. Recent Oppo 105 has a pair of superior processors. I assumed, on that basis alone, the Oppo would beat my Sony Carousel. This (presumably) single processor Sony surprisingly sounds as good as both the Onkyo and Oppo. Thus, the l/r time difference of a single processor may be less than half a millisecond?

3. LP’s, analog. After nearly 50 years of attentive listening to Reel to Reel, 8 Tracks, Cassettes, home made cassettes made from LPs, LP’s, CDs and SACD:

My simple summation is that Analog ’Gets the Overtones Right’. Perfect time alignment to the fundamentals, perfect volume decay. Chopped Digital Re-construction is an amazing achievement, but many, my friends and I included find LP’s preferable to the CD version. I’ve cooked up ’somewhat blind, out of sight switching’ in the past, no matter what: everyone chooses LP over CD and Tube over SS.

If you've played Dance Dance Revolution or other rhythm games then you quickly realize small delays make a huge difference. :-)

But yes, I also agree it makes a huge difference in the quality of audio playback to ensure your speakers are time-aligned relative to each other, and also that the difference frequencies reproduced by a single speaker are time-aligned (as much as possible).

That means using hardware components or software-based correction that facilitates that, and optimizing relative phase for turntable setups.
Tannoy single driver designs produce a singular source of directionality, this is their design philosophy, and many like yourself like/prefer the results.

however, they are not time aligned.

high frequencies travel faster than low frequencies. time aligned speakers have the front faces slanted, or separate enclosure(s) for the upper mids and perhaps a separate enclosure for the tweeters, some super-tweeters. Those separate enclosures are staggered rearward from the face of the woofer and lower mids.

All this to have the highs ’start’ further back, so that at typical listening distances, they arrive ’time aligned’.

My JSE Infinite Slope Model 2’s were the most accurate speakers I owned. Aside from the steep crossover, excellent driver choices, ... the front panel was slanted for time alignment.

My current speakers are flat faced, I put a 1-1/2" block under the front edge, tilting the entire enclosure ’back’ for 3 reasons.

1. time alignment

2. tweeters dispersion becomes directly at seated ear height

3. angle of dispersion of all drivers, especially bass to floor, and everything off the rear wall is significantly altered, the complex mixture of reflections is mixed/weakened relative to the initial sound waves.

tilting a single driver can get the 2nd and 3rd benefits, but time alignment will not/cannot be achieved. It will not alter the fixed time alignment.
Hey guys, .... I don't want to bust your bubble, .... but most speakers, even high end brands, ... are not time aligned.  Based on articles I have read, Vandies and Green Mountain Audio are supposed to be time aligned.  I believe both brands use a recessed tweeter and 1st order crossovers.  Roy Johnson of Green Mountain Audio wrote some great articles that explain why higher order crossovers screw up time coherence.  You'd have to find the articles to understand the science.

The problem of using 1st order crossovers is that the various drivers may be operating outside their optimal frequency range.  So special drivers must be used to achieve good results.  

In any case, most of us are happy with our speakers, even if not time coherent. 

If you want to see bench tests of your favorite speakers, check out John Atkinson (Stereophile) bench test results.  For example, JA measured the Vandy Model 7 MK II here: www DOT


The Revel Ultima Salon II here: www DOT 

And who doesn't like the Revel Ultima Salon II?? ;)