Topic: Matrix recordings and delays (long!)

[dklein]

Since getting my UA-5, most of the recording I've been doing has been board/mic mixes.  I've got the hang of setting the levels where I like and have been quite pleased with the results...but still have some questions about the delay effect from the mics.  

My ears tell me I'm OK but I don't have anything to compare and would like to understand the theory behind it.  I should probably also state that I am not in pursuit of the ultimate 'stereo image' (which from a phase perspective becomes impossible as soon as you mix another source in).  In fact, I don't care at all - I enjoy listening to the tonal balance, instrument mix, clarity and overall feel that seems to come from mixing mic/sbd sources.  

I understand the best thing to do is use an onstage mic to minimize/eliminate delay but that's not always practical.  Alternately, I could run a delay device on the board feed but I'm not sure I want to get into that (extra gear, extra $, additional signal processing).

I'm usually recording in small clubs - mics are 20-30 feet from PA.  Here's what I think I know about delay...

1. Sound travels at ~ 1 ms/ft so a 20 foot distance = 20 ms delay

2. The Haas effect says that until a delay exceeds 20-30ms, our brains will blend the sounds together.  Beyond that, our brains will label the first arriving sound as the source.

Now as far as I can tell, the Haas efect is more about sound localization than it is about precision...a psychoaccoustic effect mostly about how 2 different points in space can blend into one without enough time delay to distinguish them.

I did a little experiment where I took 2 copies of a song from a CD and multitracked them with different delays.  I found that a very small time shift made it really ugly sounding - even 1/2 ms.   It was interesting to hear that a big time shift like 20ms sounded much better than a small one - kind of gave it some space and stereo feel as opposed 'tunnel' sound.  

I'm not sure this really tells me much as the 2 sources were identical (unlike a board and mic source).  I can see why 2 identical sources would create phase problems when one is slightly delayed, whereas 2 different sources will not be as 'perfectly' unmatched, if that makes sense.

Now you may say just record them separately and mix later.  Definite advantages there, especially if the sound guy changes your board levels through the show (and as a result, your mix balance).  But then I need 2 rigs (I suppose I could do the board on my minidisc).  Also, a couple of times in the past I've aligned 2 sources and have found it difficult to get an exact lock on the position.  The variance between two sources has been as much as +/- 10 ms depending on where I synch (like source 1 is ahead of source 2 at one point, behind at another, then ahead again after that).  I usually ended up settling on an average and the sound didn't seem to suffer within those parameters.  This +/- variance occurs even if I match the start and end points of a recording by stretching one.  I believe this variance is there because the mic source contains so much reflected sound that it really depends what I choose as a synch point.  I usually go for a well defined peak in the waveform - a big kick drum peak or a hi-hat peak.  Since different frequencies (like kick drums and hi-hats) have different reflective characteristics, I think this is the source of the +/- variance.
I'm not talking about time drift here - this variances can occur within a few seconds of each other.  Then there's the whole time drift issue from different ADCs but let's ignore that for the moment and assume that both sources are the same lenth.

Tonight I went back to a 2 source mix and started moving one source back and forth.  Seems like I could get away with +/- 25 ms before anything sounded terribly different.  hmmmm....kinda similar to that Haas number.

Some questions  
Can anyone shed some additional technical light on the time/distance thing?
Is a matrix such a phasey mess that you can just say anything under 30 feet is OK? Or for that matter, is a mic in a bar so filled with reflected sound that it doesn't matter?
Is the balance between board/mics a factor? - I definitely go heavier on the board
How are 'professional' live recordings done? (yeah that's a loaded question)

david

p.s. here's a sample track that had 20-25 ft to the mics mixed live with the board http://tinyurl.com/j64t

[the magoo]

ok bro these are just publications....if u can find them...wich in the web im sure u can.....these would be very usefull here u go personal.riverusers.com/~manderso/uhjdisc/ambipubl.htm#m good luck bro...


Goo......there is all kinds of answers if u can find them 8)edit:in which now that ive tried is pretty damn tough....good luck...ill leave it up for u to try though bro

[nic]

ah, the matrix questions.

1 thing you need to bear in mind is that the speed of sound changes in relation to room temperature and humidity.
also, the 1 foot = 1ms isnt a rule...its a general theory.

I work with a band where we create matrix recordings and there have been venues where the distance from the PA to the SBD(where the ambient mics were at) as 40 feet but the delay needed for the recording was only 20ms.

in our experience, the best easiest way to make a matrix with only a 2 channel recording is to either run the mics on-stage and feed back to the SBD through the snake, or run the SBD feed on the snake return up to the stage to mix with on-stage mics.
if you cant do the above then trey to make a matrix only when the venue size is small...ie, the distance from PA to SBD is less than 30feet.
of course if you can multitrack(record the mics and SBD feed as seperate tracks) you can always add the delay in post production.

as for mixing, Im in the camp that prescribes to the theory of 60/40(sbd/mics)...unless the mics pick up too much chatter from the crowd

[dklein]

Thanks - understood that the speed changes with temperature and humidity but in theory the change is quite small - especially when we're talking about 25 ft.

Here's an extreme scenario - start at 20 degrees celsius, 20% humidity and go to 30 / 90.  Delay changes from 22.8 ms to 23.3 ms for an 8 metre distance (about 25 feet).  Pretty much nothing because of the short distance.  For you non-metric guys - that goes from cool and dry to hot and steamy!

http://www.measure.demon.co.uk/Acoustics_Software/speed.html

Is the reality that much different than the theory?  Maybe signal processing equipment / amps that come after the board feed create further delays on the p.a.?

How did you decide on the delay - was it live (by remote monitoring?) or post production?

[nic]

I always do mixing/delay post production, but we record multi-track with a MOTU828> DP3

[dklein]

Alright..am I in too deep here - thought I'd bump it up and see if anyone's got anything to add.  Inquiring minds want to know  

[rabhan]

yeah, i was wondering the same thing with using the ua5. to get a 30/70, 40/60, or 50/50 mix, how do i set the inputs? should i have the sbd signal feeding to like -20 and have my mics fill from there or what?